Methodology: equations, sources and test cases
How every number on this site is produced: the equations we compute, the ones we chose not to, the reference tables, and the test cases the engine must pass before anything is published.
Sources checked November 21, 2026.
At a glance
- Equations computed
- 11
- Test cases
- 52, all passing
- Sources checked
- November 21, 2026
On this page
How a result is produced
Each calculator applies one published equation to your inputs in full precision, then rounds once for display, half-up to one decimal place for VO2 max. Weights entered in pounds are converted to kilograms (and back to pounds for the Rockport equation, which was fitted in pounds) at 2.20462262 pounds per kilogram, and miles to meters at 1,609.344. Times are read as minutes and seconds, so 14:30 means fourteen and a half minutes. Any difference shown on a page is worked out from the rounded figures you see. The same engine runs in your browser and when the pages are built, so the worked examples always match what the calculator shows.
The typical error shown under a result is plus or minus one standard error of estimate from the paper, using the check-group figure where the paper gives one. Where no standard error was published, the result says so. Percentiles use straight-line steps between the published FRIEND points; below the 5th or above the 95th percentile they are given in words.
Equations checked against the original papers
On November 21, 2026 every equation that had only been seen in secondary sources was checked against its original paper: the full text where we could read it, otherwise the paper's own abstract on the publisher's site, PubMed Central or Europe PMC. Pages that refused automated reading were not worked around. An equation we could not confirm is not computed anywhere on the site.
Three sets of equations are the exception, and are labeled as such: Cooper's 12-minute conversion, the ACSM walking, running and cycling equations, and the Daniels-Gilbert running equations. Cooper's abstract confirms his group and correlation but prints no equation, so we use the conversion printed in later peer-reviewed studies that cite him. The ACSM equations come from a textbook we have not read; we use the form printed by the Compendium of Physical Activities and, for walking and running, in a 2020 peer-reviewed study. The Daniels-Gilbert equations come from a 1979 book we have not read; we use the form reproduced in later sources and check it against the publisher's own excerpt of Daniels' later book. The Checked line under each formula below says this too.
| Equation | Paper | Result | Read | What the site does |
|---|---|---|---|---|
| Age-predicted maximum heart rate, 208 - 0.7 x age | Tanaka et al. 2001 | Confirmed | Full paper | Used as published |
| Rockport one-mile walk | Kline et al. 1987 | Confirmed | Paper's published abstract | The site uses the L/min equation printed in the abstract and divides by body mass. The mL/kg/min form in secondary sources is not printed there and is not used. |
| 1.5-mile run | George et al. 1993 | Not confirmed: not computed | Abstract gives statistics only | Held. The same paper's 1-mile jog equation is printed and is used instead. |
| Non-exercise questionnaire equation | Jackson et al. 1990 | Not confirmed: not computed | Abstract gives statistics only | Held. Described, not computed. |
| Queens College step test | McArdle et al. 1972 | Not confirmed: not computed | No abstract or full text could be read | Held. Described, not computed. |
| 20 m shuttle run (beep test) | Léger et al. 1988 | Confirmed | Paper's published abstract | The site uses the age equation printed in the paper, with age held at 18 for adults. A different adult form seen in secondary sources is not used. |
| Bruce treadmill time equation | Foster et al. 1984 | Not confirmed: not computed | No abstract or full text could be read | Held. |
| Single-stage treadmill walking test | Ebbeling et al. 1991 | Confirmed | Paper's published abstract | Used as published |
| Peak cycling power to VO2 max | Hawley and Noakes 1992 | Not confirmed: not computed | Abstract gives the correlation only | Held. The finding is described without an equation. |
| Heart rate ratio factor for women | Uth et al. 2005 | Confirmed | Paper's published abstract | Used as published |
Equations computed on this site
How this is calculated: Heart rate ratio
VO2 max = 15.3 × HRmax ÷ HRrest for men, 14.5 × HRmax ÷ HRrest for women
- Source
- Uth et al. 2004; Uth et al. 2005
- Built on
- 46 well-trained Danish men aged 21 to 51 (the men's factor came from a subgroup of 10), and 27 trained women in the 2005 study
- Reported error
- Standard error about 4.5 percent (2.7 mL/kg/min) in men with a measured maximum heart rate, about 7.8 percent (4.7) with an age-predicted maximum, and about 6.8 percent in women
- Checked
- November 21, 2026, checked against the paper's published abstract
How this is calculated: Age-predicted maximum heart rate
HRmax = 208 - 0.7 × age
- Source
- Tanaka et al. 2001
- Built on
- A meta-analysis of 351 studies (18,712 people), then a lab check in 514 healthy adults
- Reported error
- Individual maximums still vary around the line, so a measured maximum is better
- Checked
- November 21, 2026, checked against the full paper
How this is calculated: Rockport one-mile walk
VO2 max (L/min) = 6.9652 + 0.0091 × weight in pounds - 0.0257 × age + 0.5955 × sex - 0.2240 × walk time in minutes - 0.0115 × heart rate; then × 1,000 ÷ weight in kg (sex is 1 for male, 0 for female)
- Source
- Kline et al. 1987
- Built on
- 343 healthy adults aged 30 to 69 (165 men, 178 women), with VO2 max measured on a treadmill; the equation came from 174 of them
- Reported error
- Standard error 0.325 L/min (r 0.93) when built, and 0.355 L/min (r 0.92) in a separate check group of 169
- Checked
- November 21, 2026, checked against the paper's published abstract
How this is calculated: Cooper 12-minute run
VO2 max = (meters covered - 504.9) ÷ 44.73
- Source
- Cooper 1968; Campos et al. 2017
- Built on
- 115 US Air Force men
- Reported error
- Correlation of 0.897 with treadmill tests; no standard error was published
- Checked
- November 21, 2026, Cooper's abstract confirms the group and correlation but prints no equation; this conversion is the form printed in later peer-reviewed studies that cite the 1968 paper, which we could not read in full
How this is calculated: 20 m shuttle run (beep test)
VO2 max = 31.025 + 3.238 × S - 3.248 × A + 0.1536 × A × S, where S = 8 + 0.5 × stage (km/h) and A is age in years, held at 18 for adults
- Source
- Léger et al. 1988
- Built on
- 188 boys and girls aged 8 to 19; checked in 77 men and women aged 18 to 50
- Reported error
- Correlation 0.71 for children and 0.90 for adults; the readable part of the abstract gives no standard error
- Checked
- November 21, 2026, checked against the paper's published abstract
How this is calculated: 1-mile jog
VO2 max = 100.5 + 8.344 × sex - 0.1636 × weight in kg - 1.438 × jog time in minutes - 0.1928 × heart rate (sex is 1 for male, 0 for female)
- Source
- George et al. 1993
- Built on
- 149 relatively fit college students aged 18 to 29 (88 men, 61 women); the jog equation came from 54 of them
- Reported error
- Standard error 3.0 mL/kg/min when built and 3.1 in a separate check group of 52
- Checked
- November 21, 2026, checked against the paper's published abstract
How this is calculated: Single-stage treadmill walk
VO2 max = 15.1 + 21.8 × speed in mph - 0.327 × heart rate - 0.263 × speed × age + 0.00504 × heart rate × age + 5.98 × sex (sex is 1 for male, 0 for female)
- Source
- Ebbeling et al. 1991
- Built on
- 139 adults aged 20 to 59; the equation came from 117 of them, walking for 4 minutes at 5 percent grade
- Reported error
- Standard error 4.85 mL/kg/min; in a check group of 22, 90.9 percent were within 5 mL/kg/min
- Checked
- November 21, 2026, checked against the paper's published abstract
How this is calculated: Daniels-Gilbert running VO2 (VDOT)
Oxygen cost = -4.60 + 0.182258 × v + 0.000104 × v² (v in meters a minute); share of maximum = 0.8 + 0.1894393 × e^(-0.012778 × t) + 0.2989558 × e^(-0.1932605 × t) (t in minutes); VDOT = oxygen cost ÷ share
- Source
- Daniels and Gilbert 1979
- Built on
- Race performances of distance runners; the original 1979 data were not read
- Reported error
- A performance index, not a measured VO2 max; no error range applies
- Checked
- November 21, 2026, the 1979 book was not read; equations as reproduced in later secondary sources, which agree with each other, and checked against the publisher's excerpt of Daniels' Running Formula (17:03 for 5 km at VDOT 60, which these equations give)
How this is calculated: ACSM walking equation
VO2 = 0.1 × speed + 1.8 × speed × grade + 3.5 (speed in meters a minute, grade as a fraction)
- Built on
- Meant for 50 to 100 meters a minute
- Reported error
- Gives the oxygen cost of a steady workload, not a maximum
- Checked
- November 21, 2026, the ACSM book itself was not read; equations as printed by the Compendium of Physical Activities and, for walking and running, in a 2020 peer-reviewed study
How this is calculated: ACSM running equation
VO2 = 0.2 × speed + 0.9 × speed × grade + 3.5 (speed in meters a minute, grade as a fraction)
- Built on
- Meant for speeds above 134 meters a minute
- Reported error
- Gives the oxygen cost of a steady workload, not a maximum
- Checked
- November 21, 2026, the ACSM book itself was not read; equations as printed by the Compendium of Physical Activities and, for walking and running, in a 2020 peer-reviewed study
How this is calculated: ACSM leg cycling equation
VO2 = 1.8 × (watts × 6.12) ÷ body weight in kg + 7
- Source
- ACSM metabolic equations
- Built on
- Steady cycling on an ergometer
- Reported error
- Gives the oxygen cost of a steady workload, not a maximum
- Checked
- November 21, 2026, the ACSM book itself was not read; equations as printed by the Compendium of Physical Activities and, for walking and running, in a 2020 peer-reviewed study
Reference tables
Percentiles come from FRIEND 2015, measured treadmill tests from US labs. Means for comparison come from HUNT 2013, healthy Norwegian adults. We read both papers' tables directly. The 2022 FRIEND update could not be read in full, so it is described but not used. The Cooper Institute ratings used by some watches are described but not reproduced. Calculations never mix tables in one rating.
Test cases
The engine must pass every case below before the site is built. Labels say plainly where each expected value came from: a value printed in a source, or a case we constructed by applying the published equation by hand (which proves the code matches the equation, not that the equation is right). Held cases check that equations we could not confirm are refused.
| Case | Type | Where the expected value comes from | Result |
|---|---|---|---|
| UT01 | Worked example printed on another calculator site | Omni Calculator VO2 max FAQ: 187 / 60 x 15.3 = about 47.7 | Pass |
| DV01 | Published check value | FitCraft VDOT page: share curve gives 88.8% at 60 minutes | Pass |
| DV02 | Published table value | Daniels Running Formula table 3.1, VDOT 50 row (table row as reproduced online): 5K 19:57 | Pass |
| DV03 | Published table value | Daniels Running Formula table 3.1, VDOT 50 row: 10K 41:21 (table row as reproduced online) | Pass |
| DV04 | Published table value | Daniels Running Formula table 3.1, VDOT 50 row: marathon 3:10:49 (table row as reproduced online) | Pass |
| DV05 | Published table value | Daniels table VDOT 50 5K 19:57; equation inverse gives 19:56 | Pass |
| DV06 | Published table value | Daniels table VDOT 50 marathon 3:10:49; equation inverse 3:10:40 | Pass |
| NM01 | Published table value | FRIEND 2015 Table 3: men 40-49 50th percentile 37.8 | Pass |
| NM02 | Published table value | FRIEND 2015 Table 3: women 20-29 75th percentile 44.7 | Pass |
| NM03 | Published table value | FRIEND 2015 Table 3: men 60-69 5th percentile 17.4 | Pass |
| NM04 | Our case built from a published table | FRIEND 2015 women 50-59: 50th 23.4, 75th 27.6; midpoint 25.5 interpolates to 62.5, shown as 63 (half-up) | Pass |
| NM05 | Published table value | Loe et al. 2013 Table 2, men 30-39 | Pass |
| CO01 | Our constructed case | Cooper 1968 equation as (d - 504.9) / 44.73 | Pass |
| CO02 | Our constructed case | Cooper 1968 | Pass |
| CO03 | Our constructed case | Cooper 1968; 1 mile = 1609.344 m | Pass |
| CO04 | Behavior rule | Equation goes negative below 504.9 m: engine must refuse, not show a negative number | Pass |
| RW01 | Our constructed case | Kline 1987 abstract equation (L/min), divided by body mass: 6.9652 + 0.0091 x 180 - 0.0257 x 45 + 0.5955 - 0.2240 x 14 - 0.0115 x 120 = 3.5262 L/min; 3526.2 / 81.6466 kg = 43.19 | Pass |
| RW02 | Our constructed case | Kline 1987 abstract equation (L/min), divided by body mass; kg converted to lb at 2.20462262 | Pass |
| RW03 | Our constructed case | Kline 1987 abstract equation (L/min), divided by body mass; tests mm:ss to decimal minutes (17:30 = 17.5) | Pass |
| RW04 | Behavior rule | Kline 1987 sample aged 30 to 69: show result with a plain warning | Pass |
| UT02 | Our constructed case | Uth 2004, factor 15.3 | Pass |
| UT03 | Our constructed case | Uth 2005 abstract: factor 14.5 for women | Pass |
| UT04 | Our constructed case | Tanaka 2001 (full text): HRmax 208 - 0.7 x 50 = 173; Uth 2004 factor 15.3 | Pass |
| GE01 | Held equation: must be refused | George 1993: 1.5-mile coefficients are not printed in the abstract and the full text could not be read. Not computed on this site | Pass |
| GE02 | Held equation: must be refused | George 1993: 1.5-mile coefficients are not printed in the abstract and the full text could not be read. Not computed on this site | Pass |
| JA01 | Held equation: must be refused | Jackson 1990: coefficients not printed in the abstract; full text not accessible. Not computed on this site | Pass |
| JA02 | Held equation: must be refused | Jackson 1990: coefficients not printed in the abstract; full text not accessible. Not computed on this site | Pass |
| QS01 | Held equation: must be refused | McArdle 1972: no abstract or full text could be read. Not computed on this site | Pass |
| QS02 | Held equation: must be refused | McArdle 1972: no abstract or full text could be read. Not computed on this site | Pass |
| BT01 | Our constructed case | Leger 1988 abstract: X = 8 + 0.5 x 10 = 13 km/h; adult age held at 18: 31.025 + 3.238 x 13 - 3.248 x 18 + 0.1536 x 18 x 13 = 50.5974 | Pass |
| AC01 | Our constructed case (textbook equation) | ACSM walking equation | Pass |
| AC02 | Our constructed case (textbook equation) | ACSM running equation | Pass |
| AC03 | Our constructed case (textbook equation) | ACSM leg cycle equation, 1 W = 6.12 kgm/min | Pass |
| UN01 | Unit definition | 1 MET = 3.5 mL/kg/min | Pass |
| UN02 | Unit definition | L/min = mL/kg/min x kg / 1000 | Pass |
| RD01 | Rounding rule | House rounding rule: differences computed from the rounded figures shown | Pass |
| TN01 | Published equation | Tanaka 2001 full text: 208 - 0.7 x age | Pass |
| UT05 | Our constructed case | Uth 2005 factor 14.5 x 180 / 65 = 40.1538 | Pass |
| BT02 | Our constructed case | Leger 1988 child form with age 12: 31.025 + 3.238 x 11 - 3.248 x 12 + 0.1536 x 12 x 11 = 47.9422 | Pass |
| BT03 | Behavior rule | Leger 1988 derived on ages 8 to 19 and adults: younger children are refused | Pass |
| BT04 | Published table value | Leger 1988 abstract: starting speed 8.5 km/h | Pass |
| GJ01 | Our constructed case | George 1993 abstract, 1-mile jog: 100.5 + 8.344 - 0.1636 x 77.111 - 1.438 x 9.5 - 0.1928 x 150 = 53.6477 | Pass |
| GJ02 | Our constructed case | George 1993 abstract: 100.5 - 0.1636 x 60 - 1.438 x 10.25 - 0.1928 x 160 = 45.0965 | Pass |
| GJ03 | Behavior rule | George 1993: jog time at least 8.0 min for men, 9.0 for women, HR at most 180, to keep the jog submaximal | Pass |
| EB01 | Our constructed case | Ebbeling 1991 abstract: 15.1 + 21.8 x 3.5 - 0.327 x 130 - 0.263 x 3.5 x 40 + 0.00504 x 130 x 40 + 5.98 = 44.258 | Pass |
| EB02 | Our constructed case | Ebbeling 1991; km/h converted to mph at 1.609344 | Pass |
| BR01 | Held equation: must be refused | Foster 1984: no abstract or full text could be read. Not computed on this site | Pass |
| HW01 | Held equation: must be refused | Hawley 1992: the abstract gives r 0.97 but not the equation. Not computed on this site | Pass |
Sources and recheck dates
Each source has a recheck interval: watch and wearable material every 90 days, training and health summaries every 180 days, reference tables and published equations every 365 days. The build refuses to publish when a source is badly overdue.