You have a creatinine result and you want the eGFR that goes with it. That calculation needs three things: your creatinine in mg/dL, your age in years, and your sex. Feed those into the CKD-EPI 2021 equation and you get a number in mL/min/1.73m². This page does the arithmetic out loud, twice, with real values, so you can follow every multiplication rather than trusting a black box.
The short answer, before the detail. Take your serum creatinine in mg/dL (divide by 88.4 first if your lab reports µmol/L). Divide it by 0.7 if you are female or 0.9 if you are male. Raise that ratio to one exponent if it is below 1 and a different exponent if it is above 1. Multiply by 142, multiply by 0.9938 raised to the power of your age, and multiply by 1.012 if you are female. That result is your estimated GFR. Everything else on this page is either explaining why each of those steps exists or showing you the numbers moving.
Two things worth saying at the start. First, this is an estimate built from a population, not a measurement of your kidneys, and the letter “e” in eGFR is doing real work. Second, there is more than one equation, they disagree with each other, and which one your report used depends on your laboratory and your country. If your question is which equation to trust rather than how to run one, Cockcroft-Gault versus MDRD covers the choice directly and this page will not repeat it. If you want the underlying concept rather than the arithmetic, creatinine clearance versus GFR explains why those two numbers are not the same thing.
The Waldev creatinine clearance and GFR calculator runs all of these equations on your values in one step, handles the unit conversion, and shows the results side by side. Read on if you want to understand what it is doing, or want to check its working.
On this page
What you need before you start
The modern creatinine equations are deliberately short on inputs. That was a design decision, not an oversight. Every extra variable an equation demands is another variable that has to be recorded accurately in a busy clinic, and weight and height are recorded far less reliably than date of birth. So the standard equation asks for three things and nothing else.
Serum creatinine, with its units. The number itself is useless without knowing whether it is mg/dL or µmol/L. A creatinine of 1.1 is unremarkable; a creatinine of 1.1 read as µmol/L would be biologically impossible. Check the unit printed beside the value on your report before you do anything else. If you are not sure what the value represents at all, creatinine in a blood test explains what is being measured.
Age in years. Whole years is fine. The age term changes the answer by roughly six-tenths of one percent per year, so a birthday next week will not move your result meaningfully. These equations were derived and validated in adults, and none of them should be applied to children, who have their own paediatric equations built on height.
Sex. The equations use a binary male or female term, which reflects the datasets they were built from rather than any claim about biology being binary. For most people this is straightforward. Where it is not, the sensible approach is to calculate both and treat the pair as a range, and to discuss the result with a clinician who knows your history rather than relying on either figure alone.
Nothing else, for CKD-EPI. No weight, no height, no ethnicity, no urine collection. If a calculator asks you for weight before it will give you an eGFR, it is either running Cockcroft-Gault or it is de-indexing the answer for body size. Both are legitimate; they are just not the same output.
What you also need is a creatinine value that means something. A result taken during a vomiting illness, in the twenty-four hours after a hard gym session, or the morning after a large steak, will produce an eGFR that describes that day and not your baseline. Dehydration raising creatinine is the single most common reason a one-off eGFR looks worse than the person actually is, and creatine supplementation is the most common reason it looks worse in someone with entirely healthy kidneys. Garbage in, garbage out applies to this arithmetic exactly as it does to any other.
One eGFR is a snapshot. Chronic kidney disease is defined by abnormality lasting at least three months, which is why a single low estimate is normally repeated rather than acted on. If you are trying to work out whether a number is worrying, when to worry about creatinine levels is the more useful page.
Getting your creatinine into the right units
This is where most hand calculations go wrong, and it goes wrong silently. All three of the classic equations were built on creatinine measured in milligrams per decilitre. Much of the world reports in micromoles per litre. If you drop a µmol/L value straight into an equation expecting mg/dL, you will not get an error message. You will get a number, and it will be catastrophically wrong.
creatinine in mg/dL = creatinine in µmol/L ÷ 88.4
creatinine in µmol/L = creatinine in mg/dL × 88.4
The 88.4 comes from creatinine’s molecular weight, 113.12 g/mol, combined with the decilitre-to-litre and milligram-to-micromole conversions. It is a fixed constant, not an approximation, and it is worth committing to memory if you look at kidney results often. Some sources round it to 88; the difference is under half a percent and will not change a clinical decision, but use 88.4 if you want your answer to match a calculator’s.
| µmol/L | mg/dL | Roughly who this is |
|---|---|---|
| 44 | 0.50 | Low end; small-framed adult, low muscle mass, or pregnancy |
| 60 | 0.68 | Comfortably normal for most women |
| 80 | 0.90 | Comfortably normal for most men |
| 97 | 1.10 | Upper-normal in a man, mildly raised in a woman |
| 124 | 1.40 | Raised in either sex; worth investigating |
| 177 | 2.00 | Substantially reduced filtration in almost anyone |
| 354 | 4.00 | Advanced impairment |
| 530 | 6.00 | Severe; usually a nephrology inpatient |
A quick sanity check that costs nothing: adult creatinine in mg/dL almost always sits between 0.4 and 2.0, and in µmol/L between 40 and 180. If your converted figure lands outside those bands, convert again before you calculate. What normal actually looks like on a report, including why the male and female ranges differ, is covered in what a normal creatinine level is.
One more unit trap. Some laboratories, particularly in the United States, still report creatinine to two decimal places while others round to one. That rounding matters more than you would expect at good kidney function, for reasons the section on why your answer differs from the lab’s comes back to. Use the value exactly as printed, decimals and all.
The CKD-EPI 2021 creatinine equation, term by term
This is the current standard for reporting eGFR from creatinine in adults, recommended by the National Kidney Foundation and the American Society of Nephrology and adopted across most of the United States, with uptake elsewhere varying by country and laboratory. Here it is in full.
eGFR = 142 × min(Scr ÷ κ, 1)α × max(Scr ÷ κ, 1)−1.200 × 0.9938Age × 1.012 [if female]
κ = 0.7 for females, 0.9 for males
α = −0.241 for females, −0.302 for males
Scr = serum creatinine in mg/dL · Age in years · result in mL/min/1.73m²
It looks worse than it is. Six terms, and each one does a job you can describe in a sentence.
142 — the scaling constant
A fitted number with no biological meaning on its own. It sets the overall scale so that the rest of the equation lands in the right range. Change nothing else and this alone determines whether a healthy young adult comes out near 100 or near 200.
κ (kappa) — the sex-specific pivot
0.7 for women, 0.9 for men. Dividing creatinine by kappa converts your absolute value into a value relative to a typical person of your sex. It is the term that accounts for men carrying more muscle and therefore more creatinine at identical kidney function.
min(Scr/κ, 1) and max(Scr/κ, 1)
The two-branch switch. If your ratio is below 1, the min term equals the ratio and the max term equals 1. If the ratio is above 1, the min term equals 1 and the max term equals the ratio. Exactly one branch is ever active; the other is 1 raised to a power, which is 1.
α (alpha) and −1.200 — the two slopes
The gentle exponent (−0.241 or −0.302) applies below the pivot; the steep −1.200 applies above it. This is why the relationship between creatinine and eGFR is not a straight line, and why a small creatinine rise matters more when creatinine is low.
0.9938Age — the age decay
Multiplies the result down by about 0.62 percent for every year of age. At 40 this factor is roughly 0.78; at 80, roughly 0.61. It encodes the observed fall in filtration with age in the populations the equation was built from.
1.012 — the residual female term
A 1.2 percent uplift applied only to women, after kappa and alpha have already done the heavy lifting. It is small because most of the sex difference is handled earlier in the equation.
The two-branch switch, in plain terms
The min and max construction confuses more people than the rest of the equation combined, so it is worth one more pass. The equation models the creatinine-to-GFR relationship as two different curves joined at a hinge, and the hinge sits exactly at the point where your creatinine equals kappa. For a man, that hinge is a creatinine of 0.9 mg/dL. For a woman, 0.7 mg/dL.
If your creatinine is below kappa (under 0.9 in a man, under 0.7 in a woman), the ratio is less than 1. The min term is your ratio, raised to the gentle alpha exponent. The max term is 1, and 1 to any power is 1, so it drops out. Only the shallow curve is in play.
If your creatinine is above kappa, the ratio is greater than 1. Now the min term is 1 and vanishes, and the max term is your ratio raised to −1.200. The steep curve takes over, and eGFR falls away quickly as creatinine climbs.
If your creatinine equals kappa exactly, both terms are 1 and both drop out. The equation collapses to 142 × 0.9938Age, times 1.012 for a woman. A man of 50 with a creatinine of exactly 0.9 gets 142 × 0.993850, which is about 104.
Why two curves? Because a single power curve fitted across the whole range gets the healthy end badly wrong. At high filtration, creatinine barely moves as GFR changes, so the equation needs a shallow slope there. At low filtration, small creatinine changes correspond to large GFR changes, so it needs a steep one. Forcing one exponent to serve both produced systematic errors in the earlier MDRD equation, particularly in people whose kidneys were fine, and fixing that was the main motivation for CKD-EPI in the first place.
An immediate practical consequence: eGFR is far more sensitive to creatinine at good kidney function than at poor kidney function. In a 40-year-old man, creatinine moving from 0.9 to 1.0 mg/dL drops eGFR from about 111 to about 98, a fall of 13 points from one-tenth of a milligram. In a 58-year-old man, creatinine moving from 1.4 to 1.5 drops eGFR from about 58 to about 54, less than five points. The same measurement noise looks alarming at one end of the range and trivial at the other.
Why the 2021 version removed the race coefficient
The 2009 CKD-EPI equation, and the MDRD equation before it, contained a multiplier applied to Black patients. In MDRD it was 1.212; in CKD-EPI 2009 it was 1.159. Both raised the reported eGFR by roughly 16 to 21 percent for a Black patient compared with a non-Black patient with an identical creatinine, age and sex. The 2021 revision removed that term entirely, and refitted the whole equation without it. That is the single most important change between the two versions, and it is worth understanding properly because it is often described badly.
Where the coefficient came from
It was empirical. When the MDRD study cohort was analysed, self-identified Black participants had, on average, slightly higher measured creatinine at the same measured GFR than other participants. The equation developers added a coefficient because it improved the statistical fit of their model. The usual explanation offered afterwards was average differences in muscle mass, but that explanation was retrofitted to the data rather than tested, and it was never a strong one.
Why it was removed
Race is a social category, not a biological variable. Nothing about self-reported race predicts an individual’s muscle mass, diet or creatinine handling. Applying a population-level average as a fixed multiplier to individuals was mixing up two different things, and the coefficient had no mechanism behind it that could be measured in a given patient.
It is unrecordable in practice. The coefficient assumed a clean binary, Black or not Black, that does not survive contact with real people. Laboratories frequently did not know a patient’s self-identified race, and had to guess or default. A variable that cannot be reliably ascertained cannot reliably be applied.
It systematically inflated eGFR in one group. A higher reported eGFR means a later CKD diagnosis, a later nephrology referral, later access to transplant waiting lists, and a later start on medications that protect the kidney. In a 58-year-old man with a creatinine of 1.4 mg/dL, the old 2009 equation gave roughly 55 without the coefficient and roughly 64 with it. Nine points sounds abstract until you notice that thresholds for referral, transplant listing and some drug decisions sit at exactly the kind of round numbers that gap straddles.
Removing it cost very little accuracy. The refitted 2021 equation was validated against measured GFR across a large pooled dataset and performed comparably overall. It is slightly less accurate in some subgroups than the race-adjusted version was, and the task force that recommended it said so openly. The judgement was that a small, evenly distributed loss of precision was preferable to a large, unevenly distributed structural bias.
The practical effect on reported numbers is asymmetric. For patients previously assigned the coefficient, reported eGFR falls, which reclassifies some into a more advanced CKD stage overnight without anything changing in their kidneys. For everyone else, the 2021 equation reports slightly higher values than 2009 did, typically by a few points. If your eGFR appeared to change between two tests taken a year or two apart with a stable creatinine, a laboratory switching equations is a genuinely common explanation, and one worth asking about before assuming your kidneys did something.
If you are comparing an old report with a new one, look for the equation name printed near the eGFR. Reports commonly state “eGFR (CKD-EPI 2021)” or “eGFR (MDRD)”. Comparing values generated by different equations is comparing two different rulers.
Worked example one: a 58-year-old man with creatinine 1.4 mg/dL
Real numbers, every step shown. This man has had a routine blood test, his creatinine came back at 1.4 mg/dL, and the report either did not give an eGFR or gave one he wants to check.
Scr 1.4 mg/dL · age 58 · male. No conversion needed, the units are already mg/dL.
Male, so κ = 0.9 and α = −0.302.
1.4 ÷ 0.9 = 1.5556. This is above 1, so the max branch is active and the min branch becomes 1.
1.5556−1.200 = 0.5885. The min term is 1−0.302 = 1, so alpha never gets used here at all.
0.993858 = 0.6972. Fifty-eight years of 0.62 percent annual decay compounds to a reduction of just over 30 percent.
142 × 1 × 0.5885 × 0.6972 = 58.26. No female term, because this patient is male.
eGFR ≈ 58 mL/min/1.73m². Reported as 58, and in most staging systems that sits in the G3a band.
Now run the same man through the other two equations, because the differences are the point.
| Equation | Working | Result | What it is |
|---|---|---|---|
| CKD-EPI 2021 | 142 × 1.5556−1.200 × 0.993858 | 58 | eGFR, mL/min/1.73m² |
| CKD-EPI 2009 | 141 × 1.5556−1.209 × 0.99358 | 55 | eGFR, older version |
| MDRD (4-variable) | 175 × 1.4−1.154 × 58−0.203 | 52 | eGFR, mL/min/1.73m² |
| Cockcroft-Gault (82 kg) | [(140 − 58) × 82] ÷ (72 × 1.4) | 67 | Creatinine clearance, mL/min |
| CKD-EPI 2021, de-indexed | 58.3 × 2.01 ÷ 1.73 | 68 | Absolute GFR, mL/min |
Same man, same blood sample, five numbers between 52 and 68. Nothing has gone wrong. They are answering slightly different questions, and the spread between them is roughly 30 percent, which is a fair illustration of how much precision to claim for any of them. The section on body surface area indexing explains why the last two are larger, and creatinine clearance versus GFR covers why clearance systematically overshoots filtration.
Worked example two: a 71-year-old woman, creatinine 97 µmol/L
Different sex, different age, and this time the lab reported in µmol/L, so the conversion step actually matters. Her weight is 62 kg and her height 158 cm, which we will need later but not yet.
97 ÷ 88.4 = 1.0973, which rounds to 1.10 mg/dL. Skip this step and you would be feeding 97 into an equation expecting about 1, producing an eGFR near zero.
Female, so κ = 0.7 and α = −0.241.
1.10 ÷ 0.7 = 1.5714. Above 1 again, so the max branch is active. Note that her creatinine is lower than the man’s in absolute terms but her ratio is slightly higher, because the female pivot is lower.
1.5714−1.200 = 0.5814.
0.993871 = 0.6430. Thirteen years older than the previous example, and the age factor has fallen by about eight percent as a result.
142 × 0.5814 × 0.6430 = 53.08. Then × 1.012 = 53.72.
eGFR ≈ 54 mL/min/1.73m².
Her comparison table, for the same reason as before.
| Equation | Result | Note |
|---|---|---|
| CKD-EPI 2021 | 54 | The number most modern labs would print |
| CKD-EPI 2009 | 51 | Older version, no race coefficient applied |
| MDRD (4-variable) | 49 | MDRD reads lower here, as it usually does above 60 |
| Cockcroft-Gault, 62 kg | 46 | Clearance, mL/min, with the 0.85 female factor applied |
| CKD-EPI 2021, de-indexed to 1.65 m² | 51 | Absolute GFR for her actual body size |
Notice how this case inverts the pattern from the man. For him, Cockcroft-Gault gave the highest figure of the set; for her, it gives the lowest. That is not a quirk of arithmetic, it is Cockcroft-Gault responding to weight, and she is a light 71-year-old while he is a heavier 58-year-old. If she were being dosed for a renally cleared drug, that 46 might matter a great deal, and the section on why dosing uses a different equation explains why a pharmacist may reach for that number rather than the 54 on her report.
A third case: what happens when creatinine is low
Both examples so far had creatinine above kappa, so the min branch collapsed to 1 and alpha never did any work. Here is a case where the other branch is the active one, because it behaves quite differently and it is the situation most healthy people are actually in.
A 29-year-old woman, creatinine 0.6 mg/dL.
0.6 ÷ 0.7 = 0.8571. Below 1, so this time the min branch is live and the max branch becomes 1−1.200 = 1.
0.8571−0.241 = 1.0378. A negative exponent on a number below 1 gives a result slightly above 1, which nudges the answer up rather than down.
0.993829 = 0.8350.
142 × 1.0378 × 0.8350 × 1.012 = 124.6, so eGFR ≈ 125 mL/min/1.73m².
Here is the part that surprises people: many laboratories will not print 125. They will print “>90” or “>60” and nothing more. That is deliberate. The equations were derived largely in populations with reduced kidney function, and their precision at the healthy end is poor, so reporting a specific figure above 90 implies an accuracy that does not exist. A calculator that hands you 125 is being more precise than the underlying science supports, and the honest reading of that number is simply “normal filtration”.
The other thing this example shows is how flat the curve is down here. Move her creatinine from 0.6 to 0.7 and eGFR only falls to about 120. Move it from 0.6 to 0.5 and it climbs to about 130. Ten points of eGFR for a change well within normal day-to-day variation. This is the flat, shallow part of the relationship, and it is why an eGFR of 95 versus 110 tells you essentially nothing, while 35 versus 50 tells you a lot. If you are trying to interpret a value at the low-creatinine end, what low creatinine means is a better guide than the eGFR arithmetic.
The MDRD equation: the previous standard, still in circulation
MDRD stands for Modification of Diet in Renal Disease, the trial whose data produced it in 1999. For roughly a decade it was the equation behind almost every automatically reported eGFR in the English-speaking world, and it has not fully disappeared. The four-variable version, in the IDMS-traceable form used with modern standardised creatinine assays, is this.
eGFR = 175 × Scr−1.154 × Age−0.203 × 0.742 [if female] × 1.212 [if Black]
Scr in mg/dL · Age in years · result in mL/min/1.73m²
Structurally it is simpler than CKD-EPI. One creatinine term rather than two branches, a power function of age instead of an exponential decay, and flat multipliers for sex and, in the original, race. The 175 replaced an earlier 186 when creatinine assays were standardised to isotope dilution mass spectrometry; if you find a source quoting 186, it is the pre-standardisation version and should not be used with a modern result.
Where MDRD goes wrong
Its weakness is specific and well documented: it underestimates GFR at higher levels of kidney function. Take the 40-year-old man with a creatinine of 0.9 mg/dL from earlier. CKD-EPI 2021 puts him at about 111. MDRD puts him at about 94. Neither number changes anything about his health, but a system built on MDRD would have flagged a large number of people with entirely normal kidneys as being in CKD stage 2 or borderline stage 3, generating anxiety, repeat testing and referrals that achieved nothing.
Systematic underestimation above 60. The single-slope structure cannot fit both ends of the range, and MDRD was fitted mostly on people who already had reduced function. Extrapolating it upward into healthy territory is where it fails.
This is why labs capped it. Reporting “>60” rather than a figure was a direct workaround for MDRD’s unreliability above that line. Many laboratories still cap their reports out of habit even after moving to CKD-EPI.
It agrees better at low function. Down in the 15 to 45 range, MDRD and CKD-EPI produce broadly similar answers, which is why the switch changed advanced CKD management very little and changed early CKD labelling quite a lot.
It carried the race coefficient too. At 1.212, MDRD’s was larger than CKD-EPI 2009’s 1.159. Laboratories still using MDRD have generally dropped the coefficient, which means an MDRD result today may or may not be the same MDRD result as five years ago.
Where you will still meet it
In drug labels and clinical trial protocols written while MDRD was standard, and which specify eligibility or dose adjustment by MDRD eGFR. In laboratories that have not migrated, particularly outside the countries that adopted CKD-EPI 2021 quickly. In old results in your own record, which is the most likely place you will encounter it, and the reason a five-year-old eGFR is not straightforwardly comparable with this year’s. And in some national reporting standards that specify a different equation again. The full comparison of when each equation earns its place is in Cockcroft-Gault versus MDRD.
Cockcroft-Gault: not GFR, and that is the point
Published in 1976, older than either CKD-EPI or MDRD, and still in daily use for one specific job. It answers a different question from the other two, and confusing them is the most common conceptual error in this whole area.
CrCl = [(140 − Age) × Weight] ÷ (72 × Scr) × 0.85 [if female]
Age in years · Weight in kg · Scr in mg/dL · result in mL/min
Three differences from the equations above, and all three matter.
It estimates clearance, not filtration
Creatinine clearance runs higher than true GFR, because roughly 10 to 15 percent of creatinine reaches the urine by active tubular secretion rather than filtration. Clearance counts both routes; GFR counts only filtration. The gap widens as kidney function falls.
It needs weight, and which weight matters
The original used actual body weight. In obesity that inflates the result substantially, so many pharmacy protocols use ideal or adjusted body weight instead. Different weight choices can change the answer by 40 percent or more in the same patient.
It is not indexed to body surface area
Its output is plain mL/min, describing this person’s kidneys. CKD-EPI and MDRD output mL/min/1.73m², describing a standardised body. That alone makes the two families of numbers non-interchangeable.
It was built on an unstandardised assay
The 1976 derivation used creatinine measurements that read slightly higher than today’s standardised assays. Applied to a modern creatinine value it therefore tends to run a little high, on top of the secretion effect.
Worked through for our 58-year-old man at 82 kg: 140 minus 58 is 82, multiplied by his 82 kg weight is 6,724, divided by 72 times 1.4 which is 100.8, giving 66.7 mL/min. Against his CKD-EPI figure of 58, that is a 15 percent difference in the direction you would expect from a clearance estimate. For the 71-year-old woman at 62 kg: 140 minus 71 is 69, times 62 is 4,278, divided by 72 times 1.1 which is 79.2, giving 54.0, then multiplied by 0.85 for female to give 45.9 mL/min. Against her CKD-EPI 54, Cockcroft-Gault reads lower, because her low body weight pulls it down.
The 0.85 female factor is a blunt instrument. It was intended to capture lower average muscle mass per kilogram in women, applied as a flat 15 percent reduction to everyone. It does not adapt to an individual’s body composition at all, and it is one of the reasons Cockcroft-Gault is now used only where its specific job requires it. What creatinine clearance is covers the underlying measurement, and the normal creatinine clearance range gives the values to compare against.
Why a drug dose may use a different equation than your report
This confuses patients and, honestly, plenty of clinicians. Your report says eGFR 54. The pharmacist adjusting your anticoagulant is working from 46. Neither has made a mistake.
The reason is historical and regulatory rather than scientific. Drug dosing thresholds are set during clinical trials, and for decades those trials defined kidney function using Cockcroft-Gault creatinine clearance. That definition then went into the drug’s licensed label. When a label says to reduce the dose below a creatinine clearance of 30 mL/min, the number it means is the Cockcroft-Gault number, calculated the way the trial calculated it. Substituting a CKD-EPI eGFR is substituting a different quantity, measured in different units, that happens to be printed in the same ballpark.
| Situation | Which figure tends to be used | Why |
|---|---|---|
| Staging chronic kidney disease | CKD-EPI eGFR, indexed | Stage definitions are written in mL/min/1.73m² |
| Deciding on nephrology referral | CKD-EPI eGFR, indexed | Referral guidance uses the same indexed thresholds |
| Dose adjustment for many older drugs | Cockcroft-Gault clearance | The licensed label specifies it |
| Direct oral anticoagulants | Cockcroft-Gault clearance | Trial eligibility and dose-reduction criteria were defined that way |
| Chemotherapy dosing | Often measured or de-indexed absolute GFR | Narrow therapeutic window; body size must not be standardised away |
| Dosing at the extremes of body size | Absolute GFR in mL/min, de-indexed | An indexed value misrepresents a very large or very small person |
The practical consequence for you is simple: do not take the eGFR from your report and check it against a dosing threshold you found online. The threshold and your number may be expressed in different currencies. This is a pharmacist’s job precisely because it is fiddly, and the detail is covered separately in creatinine clearance and drug dosing.
What the 1.73 m² actually means, and when to remove it
Almost nobody reads the units on an eGFR properly. They are mL/min/1.73m², and that trailing figure is not decoration. It says the result has been scaled to the body surface area of a standardised adult.
The 1.73 m² figure came from a 1927 estimate of the average body surface area of a 25-year-old adult. It is genuinely that old, and it does not describe the average adult of today particularly well. It survives because it has been the reference point in every kidney dataset for decades, and changing it now would break comparability with all of them.
Why index at all
Because kidney size scales with body size. A 195 cm, 120 kg adult has larger kidneys and genuinely filters more blood per minute than a 150 cm, 45 kg adult with identical kidney health. If you compared their raw filtration rates, the smaller person would look impaired when nothing is wrong. Indexing to a common body size lets you ask “how healthy are these kidneys” rather than “how big is this person”.
Absolute GFR (mL/min) = indexed eGFR (mL/min/1.73m²) × BSA ÷ 1.73
Mosteller BSA (m²) = √( height in cm × weight in kg ÷ 3600 )
Du Bois BSA (m²) = 0.007184 × height0.725 × weight0.425
Mosteller and Du Bois usually agree to within a couple of percent, and Mosteller is far easier to do on a phone. For our 58-year-old man at 178 cm and 82 kg: 178 times 82 is 14,596, divided by 3,600 is 4.055, square root 2.01 m². His indexed eGFR of 58 becomes 58.3 times 2.01 divided by 1.73, which is 68 mL/min of actual filtration. For the 71-year-old woman at 158 cm and 62 kg, BSA works out at 1.65 m², and her indexed 54 becomes 51 mL/min absolute. He is bigger than standard so his absolute figure is higher; she is smaller so hers is lower.
De-index for drug dosing where the label calls for absolute clearance, and particularly for agents with a narrow therapeutic window such as many chemotherapy drugs. The drug is cleared by this person’s actual kidneys, not by a standardised pair.
De-index at the extremes of body size. For someone who is very large or very small, the indexed number and the absolute number diverge enough to change decisions. Above roughly 2.2 m² or below roughly 1.4 m² the difference stops being academic.
Stay indexed for staging and referral. CKD stages, referral criteria and the thresholds in most guidelines are all written in indexed units. De-indexing before comparing to a stage boundary reclassifies people incorrectly.
Stay indexed when tracking change over time, unless the person’s weight has shifted substantially, in which case the absolute figure may be the more honest one to trend.
An easy way to hold this: the indexed number answers “how are these kidneys doing, compared with everyone else’s”. The absolute number answers “how many millilitres per minute is this actual body clearing”. Staging wants the first. Drug clearance wants the second.
Why your calculated number differs from the one on your report
You have done the arithmetic carefully and got 58. The report says 61. This happens constantly and it is almost never a mistake on either side.
| Cause | Typical size of the gap | How to check |
|---|---|---|
| The lab used a different equation | Up to 15 percent, sometimes more | Look for the equation name printed beside the eGFR |
| Creatinine was rounded before you saw it | Several points, more at high eGFR | Ask for the unrounded value; recalculate with it |
| You rounded mid-calculation | 1 to 3 points | Keep four decimal places until the final step |
| The lab still applies a race coefficient | 16 to 21 percent | Rare now, but check the report footnotes |
| Your age changed between test and calculation | Under 1 point per year | Use your age on the date of the blood draw |
| Assay calibration differences between labs | Small but real | Unavoidable; a reason not to compare labs directly |
| The report capped the value | Anything above 60 or 90 | “>60” is a cap, not a result |
| The lab used cystatin C, or a combined equation | Can be large in either direction | The report will name cystatin C if it was used |
The rounding point deserves expanding, because it is the one people dismiss. A laboratory that reports creatinine as 1.0 mg/dL may have measured 0.95 or 1.04. In a 40-year-old man, those two values give eGFRs of about 104 and about 93. Eleven points of apparent difference, entirely inside the rounding of a single printed digit. That is not a flaw in your maths; it is the resolution limit of the input. It is also a good argument for treating eGFR as a band rather than a point, and for not reading anything into a change of three or four points between tests.
If your calculated figure and the report’s figure disagree by more than about 15 percent and you have checked your units and equation, the most likely remaining explanation is that the lab used a creatinine value you are not looking at, or a different equation than you assumed. Ask. Laboratories are usually happy to say which equation they run.
Where every creatinine-based equation fails
All of these equations rest on one assumption: that your creatinine production is typical for your age and sex. When it is not, the equation is solving the wrong problem, and no amount of careful arithmetic rescues it. These are the people for whom a creatinine eGFR should be treated with real caution.
Very muscular people. A 25-year-old competitive lifter with a creatinine of 1.3 mg/dL calculates to an eGFR around 78, which looks like early kidney disease. His creatinine is high because he produces a lot of it, and his filtration may be entirely normal. The equation has no muscle-mass input and cannot tell the difference. Whether creatine supplements add to this is covered in does taking creatine increase creatinine.
People with very low muscle mass. The mirror image, and clinically the more dangerous one. Frail elderly people, amputees, those with advanced liver disease, spinal cord injury, malnutrition or cachexia produce far less creatinine, so their blood level stays reassuringly low while filtration falls. An 85-year-old frail woman with a creatinine of 0.8 calculates to about 72, which sounds fine, and may substantially overstate her true filtration. This is how significant impairment goes unnoticed.
Anyone whose kidney function is changing quickly. These equations assume a steady state, where production and clearance are in balance. In acute kidney injury they are not: creatinine lags real events by a day or two, so a value taken while function is falling makes things look better than they are, and one taken during recovery makes them look worse. During an acute illness, the direction and speed of change tell you more than any calculated eGFR.
Pregnancy. Filtration rises substantially in pregnancy and creatinine falls, and none of these equations were derived or validated in pregnant women. A creatinine at the top of the ordinary range can represent significant impairment in the third trimester. Creatinine levels in pregnancy covers what actually applies.
Children and adolescents. CKD-EPI, MDRD and Cockcroft-Gault are adult equations. Paediatrics uses height-based equations instead, because a child’s creatinine tracks growth. Never run a child’s result through an adult equation.
Extremes of body size. Severe obesity and very low body weight both distort the relationship, and they distort Cockcroft-Gault worst of all because weight sits in the numerator.
Anyone on a drug that blocks tubular secretion. Trimethoprim, cimetidine, cobicistat and dolutegravir raise measured creatinine by closing the secretion route without touching filtration. The eGFR drops on paper and the kidneys are unchanged. Medications that raise creatinine lists the main offenders.
Unusual diets. A large cooked-meat meal in the previous day lifts creatinine measurably, because cooking converts meat creatine into creatinine that you absorb directly. Strict vegetarians run lower creatinine and can have their function overestimated. Neither appears anywhere in the equations.
There is also a general accuracy limit that applies to everyone, not just the special cases. When CKD-EPI 2021 was validated, roughly nine in ten estimates fell within 30 percent of measured GFR. That is considered good performance for an estimating equation. Turn it around and it means about one in ten estimates was out by more than 30 percent, and that an eGFR of 50 is realistically compatible with a true GFR somewhere in the high 30s to mid 60s. Reading a change from 52 to 49 as decline is over-reading the instrument.
Because of these limits, guidelines treat a low eGFR as a reason to look further rather than as a diagnosis. Urine albumin-to-creatinine ratio, blood pressure, imaging and the trend over months all feed into the assessment, and NIDDK sets out how those tests fit together. Calculating the albumin-to-creatinine ratio is the natural companion calculation, and the two together classify risk far better than either alone.
When cystatin C is used instead
Cystatin C is a small protein produced by essentially every nucleated cell in the body at a fairly constant rate. It is filtered by the glomerulus and then reabsorbed and broken down in the tubule, so like creatinine its blood level rises as filtration falls. Its advantage is straightforward: production does not depend on muscle mass, so it sidesteps the single biggest weakness of creatinine.
There is a CKD-EPI cystatin C equation and a combined creatinine-plus-cystatin C equation, and both follow the same structural pattern as the creatinine equation: a scaling constant, a two-branch min and max switch around a pivot value, an exponential age term, and a small sex adjustment. The combined equation uses both markers at once and is the most accurate of the three in most validation work. The exact coefficients differ between versions and it is not worth reproducing them from memory here; if you need those numbers, use the Waldev calculator or the equation as published by your laboratory, which will be the version they have implemented.
When it is worth requesting
When a creatinine eGFR sits near a decision threshold and the decision matters. When body composition is atypical in either direction. When a diagnosis of CKD would follow from a creatinine eGFR that does not fit the clinical picture.
What its own weaknesses are
Thyroid disease, corticosteroids, marked obesity, smoking and active inflammation all shift cystatin C independently of kidney function. It is not a clean marker, it is a differently dirty one.
Why it is not routine
Cost and availability, largely. The assay is more expensive than creatinine and not offered by every laboratory, so it tends to be used as a confirmatory test rather than a first-line one.
What agreement between the two means
When creatinine-based and cystatin C-based estimates agree closely, confidence in the figure goes up substantially. When they diverge sharply, that divergence is itself informative and usually prompts a closer look.
Current guidance from the major kidney bodies encourages wider use of cystatin C for confirmation, particularly where a creatinine-based eGFR is being used to make a consequential decision. If your clinical picture and your calculated eGFR do not match, it is a reasonable thing to ask about.
Mistakes people make doing this by hand
Every one of these has been made by someone confident they had it right.
The big one. A creatinine of 88 µmol/L entered as 88 mg/dL produces an eGFR of well under 1, which should be obviously wrong but often is not, because people trust the arithmetic more than their own sense of scale. Convert first, every time.
Only one branch is ever live. Multiplying the ratio by itself under two different exponents double-counts your creatinine and gives a badly wrong answer. If the ratio is above 1, the min term is 1. If below, the max term is 1.
Raising to −1.200 means dividing by the ratio to the power 1.200, not multiplying. On a phone calculator, use the xy key with the minus sign included, then check the direction: higher creatinine must give lower eGFR.
Kappa is the divisor, 0.7 or 0.9. Alpha is the exponent, −0.241 or −0.302. They belong to the same sex but do different jobs, and swapping them produces a plausible-looking number that is wrong.
Different quantity, different units, different derivation. They are not interchangeable even though both come out in the fifties and sixties for the same patient.
You cannot recalculate anything from “>60”. Go back to the creatinine, which is always printed, and calculate from that.
Round only the final answer. Rounding the ratio to two decimals before exponentiating typically costs a point or two, which is enough to move someone across a stage boundary that they were sitting on.
That constant belongs to pre-standardisation creatinine assays. With a modern IDMS-traceable result it overestimates by roughly 6 percent. Use 175.
A value taken while you are vomiting, septic or badly dehydrated describes that day. It is not your kidney function, and it usually recovers.
Staging requires abnormality persisting three months or more. A single low eGFR is a prompt to repeat the test, not a diagnosis, and treating it as one causes a lot of avoidable distress.
What to do with the number once you have it
An eGFR sits inside a staging system, but the stage is only half of the risk picture, because albuminuria carries independent weight. Broadly: 90 and above with no other abnormality is normal, 60 to 89 is mildly reduced and common with age, 45 to 59 and 30 to 44 are the two halves of stage 3, 15 to 29 is severely reduced, and under 15 is kidney failure. The detail of what each band means in practice is covered in stage 3 kidney disease, stage 4, and the levels that indicate kidney failure. If your interest is in what to do next, how to lower creatinine levels covers the evidence-based options and is honest about which ones do very little.
Get medical help urgently, whatever your calculated number says, if you have a marked drop in how much urine you are passing, new swelling of the legs, ankles or face, breathlessness at rest or lying flat, confusion or unusual drowsiness, or persistent vomiting that stops you keeping fluids down. Those symptoms need assessment now, not a calculation.
Frequently asked questions
How do you calculate GFR from creatinine?
Use the CKD-EPI 2021 equation. Convert your creatinine to mg/dL first, dividing by 88.4 if your lab reports µmol/L. Divide that by 0.7 if female or 0.9 if male. If the ratio is above 1, raise it to the power −1.200; if below 1, raise it to −0.241 for a woman or −0.302 for a man. Multiply by 142, then by 0.9938 raised to your age in years, then by 1.012 if female. The answer is in mL/min/1.73m².
Can I calculate GFR from creatinine alone?
No. Creatinine on its own is not enough, because the same value means different things in different people. A creatinine of 1.1 mg/dL is unremarkable in a 30-year-old man and represents meaningfully reduced filtration in a 75-year-old woman. Every equation needs at least age and sex alongside the creatinine, and Cockcroft-Gault also needs weight. If a calculator asks only for creatinine and gives you a GFR, it is assuming values for the rest and you should not trust the output.
What is the formula for eGFR?
The current standard is CKD-EPI 2021: eGFR equals 142, multiplied by min(Scr/κ,1) raised to α, multiplied by max(Scr/κ,1) raised to −1.200, multiplied by 0.9938 to the power of age, multiplied by 1.012 if female. Kappa is 0.7 for females and 0.9 for males; alpha is −0.241 for females and −0.302 for males. Creatinine goes in as mg/dL and the result comes out as mL/min/1.73m². Only one of the min and max terms is ever active.
Why did they remove race from the eGFR equation?
Because race is a social category with no measurable biological mechanism behind it, and applying a fixed multiplier to individuals on that basis was not defensible. The coefficient systematically raised reported eGFR in Black patients, which delayed CKD diagnosis, specialist referral and transplant listing. It was also frequently unrecordable in practice. The 2021 equation was refitted without it and validated across a large pooled dataset, accepting a small loss of precision in exchange for removing a structural bias that fell unevenly on one group.
Why is my calculated GFR different from my lab report?
Most often because the laboratory used a different equation. CKD-EPI 2021, CKD-EPI 2009 and MDRD can differ by 15 percent or more on the same sample. Other common reasons are that the creatinine you can see has been rounded before printing, which matters more than people expect at good kidney function, that you rounded partway through the calculation, or that the report capped the value at “>60” rather than reporting a figure. Check which equation is named beside the eGFR on the report.
What does the 1.73 m² in mL/min/1.73m² mean?
It means the result has been scaled to a standard adult body surface area, so kidney function can be compared between people of different sizes. The figure comes from a 1927 estimate of the average adult and survives for continuity rather than accuracy. To convert to your actual filtration in mL/min, multiply the indexed eGFR by your own body surface area and divide by 1.73. Staging and referral thresholds use the indexed number; some drug dosing uses the absolute one.
Is Cockcroft-Gault the same as eGFR?
No, and treating them as interchangeable is a genuine error. Cockcroft-Gault estimates creatinine clearance in mL/min, which runs higher than true filtration because 10 to 15 percent of creatinine reaches the urine by tubular secretion rather than filtration. It also requires weight and is not indexed to body surface area. CKD-EPI estimates glomerular filtration rate in mL/min/1.73m². They are different quantities in different units, and drug labels written around Cockcroft-Gault mean that specific number.
How accurate is an eGFR calculated from creatinine?
Reasonably good for a population, imprecise for an individual. In validation studies of CKD-EPI 2021, roughly nine in ten estimates fell within 30 percent of measured GFR, which means about one in ten was further out than that. Practically, an eGFR of 50 is compatible with a true value anywhere from the high 30s to the mid 60s. Treat it as a band rather than a point, and do not read a change of three or four points between tests as meaningful.
Which equation should I use to calculate my GFR?
For general purposes, CKD-EPI 2021, because it is the current standard and it performs better than MDRD at normal and near-normal kidney function. Use MDRD only if you are deliberately reproducing an older result or following a protocol that specifies it. Use Cockcroft-Gault only when a drug label or dosing protocol calls for creatinine clearance. Whichever you use, record which one, because comparing numbers from different equations is comparing different rulers.
Can I calculate GFR for a child using these equations?
No. CKD-EPI, MDRD and Cockcroft-Gault were derived and validated in adults, and applying them to children produces unreliable results. Paediatric estimation uses height-based equations, because a child’s creatinine tracks growth and muscle development in a way the adult equations do not model. Anyone needing a GFR estimate for a child should have it calculated by a clinician using the appropriate paediatric equation, and interpreted against age-specific reference values rather than adult ones.
The short version
Converting creatinine to GFR takes three inputs: creatinine in mg/dL, age and sex. The CKD-EPI 2021 equation divides your creatinine by 0.7 or 0.9, applies one of two exponents depending on whether that ratio lands above or below 1, scales by 142, decays by 0.9938 per year of age, and adds a 1.2 percent uplift for women. The 2021 revision dropped the race coefficient that earlier versions carried, which lowered reported values for patients who had previously been assigned it.
MDRD is the older equation and reads low at good kidney function. Cockcroft-Gault estimates creatinine clearance rather than filtration, needs weight, is not indexed to body surface area, and remains the number many drug labels mean. All of them assume your creatinine production is typical, which is why they mislead in very muscular people, frail people and anyone whose function is changing fast. Run your own figures through the Waldev creatinine clearance and GFR calculator, browse more in the creatinine blog category and the wider health blog, see the rest of the health calculators, or start from the beginning with what creatinine is and the full tool library at waldev.com.
Medical disclaimer: This article explains how estimated GFR equations work and is general educational information only. It is not medical advice, cannot tell you what your own result means, and must not be used to decide whether to seek care, delay care, or change any medication or dose. Estimated GFR is an approximation with meaningful error in individuals, and reference ranges and equations vary between laboratories. Always discuss your results with a doctor or qualified healthcare professional, and seek urgent medical attention if you have symptoms that concern you.
The National Kidney Foundation on what eGFR is, the equations behind it and how the stages are defined. Estimated GFR explained →
MedlinePlus covers what the blood test measures, how it is done and what affects the result. Creatinine test →
NIDDK on the blood and urine tests used together to assess kidney function, and why both are needed. CKD tests & diagnosis →