VDOT Explained: One Race Result, Every Training Pace
What VDOT actually measures, how it turns a single race time into five training paces, and where the number stops being true for your own running.
You ran a 5K in 22:40 three weeks ago. Now you are staring at a training plan that tells you to do "6 x 1000m at interval pace" and you have no idea what number that is. So you pick something that feels hard, run the first two reps too fast, and turn a VO2max session into a race you did not enter.
This is the problem VDOT solves. One race result, five training paces, no guessing. It is the most quietly useful number in distance running and most people who use a calculator to generate it have never been told what it actually represents.
VDOT is a performance number, not a physiology number
The temptation is to read VDOT as VO2max with a different name. It is not, and the distinction is the whole reason the system works.
VO2max is a measured rate: how much oxygen you can consume per kilogram of body weight per minute, established on a treadmill with a mask on your face. It is a ceiling. It says what your aerobic engine is capable of, and it says almost nothing about how fast you can run, because two runners with the same engine can burn wildly different amounts of oxygen holding the same pace.
That difference is running economy, and it is where most of the variance between similarly-gifted runners lives. Jack Daniels' insight, developed across decades of work with elite runners and set out in Daniels' Running Formula, was that you can sidestep the measurement problem entirely. Instead of measuring the engine and then correcting for efficiency, work backwards from the result. If somebody raced 5,000m in 22:40, whatever combination of engine and economy produced that is the combination we should be prescribing training for.
So VDOT is a pseudo VO2max: the value that reproduces your actual race performance. Daniels wrote it in capitals as VDOT to distinguish it from laboratory VO2max, and it is the more useful of the two for the simple reason that it is derived from the thing you are trying to improve.
The practical consequence is that a runner with a lab-measured VO2max of 55 might carry a VDOT of 52 or 58 depending on how efficiently they run, and the VDOT is what should drive their paces. The running zone calculator does the conversion from a race result in one step.
What the model is actually doing
Underneath, two relationships are being solved at once.
The first is the oxygen cost of running at a given velocity. For distance running on the flat, that relationship is close enough to linear across the range most people train in that it can be expressed as a simple equation, which is why you can reason about pace and oxygen demand interchangeably.
The second is how long a human can hold a given fraction of their maximum. Nobody holds 100% of VO2max for an hour. The sustainable fraction falls predictably with duration: very high for a few minutes, around 88 to 92% for a 5K depending on the athlete, closer to 80% for a half marathon, lower still across a marathon.
Put those together and a race result becomes solvable. You know the velocity, so you know the oxygen cost. You know the duration, so you know roughly what fraction of maximum was sustainable. Solve for the maximum that satisfies both, and that is your VDOT.
This also explains a property that confuses people the first time they see it: different race distances from the same fitness should produce nearly the same VDOT. A 22:40 5K, a 47:10 10K and a 1:44 half marathon are all roughly VDOT 46. If your own results disagree by more than a couple of points, that disagreement is a finding rather than an error. A 5K well above your half marathon usually means an aerobic base that has not been built out yet. A half marathon well above your 5K usually means plenty of endurance and not enough speed work.
The five paces, and what each one is for
A VDOT value expands into five training paces. The names are Daniels' and they have specific jobs.
Easy (E). Roughly 59 to 74% of VO2max, conversational, and the pace for the large majority of weekly volume. Its purpose is capillary and mitochondrial development plus structural resilience, and none of that requires speed. This is the pace runners get wrong most often, and always in the same direction.
Marathon (M). Close to your current marathon race pace, used in long runs and marathon-specific work. For runners not training for a marathon it is a useful "steady" gear that sits between easy and threshold.
Threshold (T). Roughly the pace you could hold for an hour in a race, around 83 to 88% of VO2max for most trained runners. This is the pace for tempo runs and cruise intervals, and it is the most productive single pace in distance running because it is hard enough to shift the lactate curve and easy enough to accumulate real volume at. Our tempo run guide goes into the session structures in detail.
Interval (I). Around 3K to 5K effort, roughly 95 to 100% of VO2max, in repetitions of three to five minutes. The purpose is time spent near maximum aerobic capacity, which is why the recovery between reps matters as much as the reps.
Repetition (R). Faster than interval, in short bursts of 30 to 90 seconds with generous recovery. This is not an aerobic stress at all. It is speed and economy work, and running it as a lung-busting effort defeats the point.
The single biggest error is running E too fast. It happens because easy pace feels embarrassingly slow relative to the effort a runner thinks they should be making, and because a 10K-fit runner's easy pace can be 90 seconds per kilometre slower than their threshold. The cost is not one bad session; it is that the hard sessions get run tired, which flattens the whole distribution into a permanent moderate grey zone. If you find the prescribed easy pace implausible, cross-check it against heart rate using our zone 2 heart rate guide and you will usually find the two agree and your instinct was the thing that was wrong.
Where VDOT stops being true
The model is only as good as the race you fed it, and there are four ways to feed it something misleading.
The race was not recent. Fitness moves. A result from three months and a training block ago describes a runner who no longer exists, in either direction. Anything older than about six weeks should be treated as a starting estimate rather than a prescription.
The race was not maximal. A parkrun run as part of a long run, a race you used as a workout, a half marathon where you eased off at 16km: all of them understate VDOT and all of them produce training paces that are too easy across the board.
The course was not fair. Elevation, wind, heat and a badly measured course all distort the result, and heat distorts it a lot. A hot-weather race can cost a minute or more over 10K, and our data on pace slowdown in heat shows how quickly that effect arrives. Feed a hot race into a calculator and you will train too slowly for weeks.
The pacing was poor. A race run as a fast first half and a survival second half produces a finish time below what the runner's fitness supports. The model sees only the finish time, so it inherits the pacing error.
There is a practical version of the fair-course problem that catches people every week, and it has nothing to do with fitness. The pace your watch shows you in real time is not the pace the model is talking about. GPS instantaneous pace is smoothed and noisy, drifts under tree cover and between buildings, and on a twisting course it under-measures distance in a way that makes you look slower than you are. Threshold and interval work are both prescribed at paces where a 5 second per kilometre error changes the physiological target. Run those sessions on a track, a measured loop, or off lap splits rather than off the live pace field, and treat the instantaneous number as a rough guide only.
Treadmills have the opposite failure. Belt speed calibration drifts, and a treadmill reading 12.0 kph while actually running 11.6 kph will hand you sessions that are consistently easier than prescribed. If most of your quality work is indoors, verify the belt against a known outdoor effort once a season rather than assuming the console.
There is a fifth limitation that is not an input problem at all. VDOT is a single number and running is not a single quality. A runner with a very high aerobic ceiling and no top-end speed, and a runner with the opposite profile, can arrive at the same VDOT and should not necessarily train the same way. The paces are a strong default. They are not a diagnosis.
VDOT alongside the other threshold models
Runners who also cycle or swim run into this quickly: every sport has invented its own version of the same idea, and they are more alike than the vocabulary suggests.
Cycling has FTP, the power you can hold for roughly an hour, and critical power, a mathematically cleaner construct derived from the power-duration curve. Swimming has critical swim speed, derived from two time trials at different distances. Running has VDOT and, separately, critical speed.
All of them are answering the same question: what intensity sits at the boundary between sustainable and not, and what should training around that boundary look like. The FTP versus critical power guide covers where those two models diverge, and the reasoning transfers directly. VDOT's distinguishing feature is that it is anchored to a race rather than to a test, which makes it easier to keep current for anyone who races regularly and harder to keep current for anyone who does not.
If you want to compare training load across sports rather than just pace within one, that is a different number again and belongs to training stress score.
Keeping the number current without testing constantly
The temptation once you have a VDOT is to chase it. Resist that, because a time trial every three weeks is a training block made of tests.
Recalculate after every genuine race. Between races, leave it alone and let the training do its work. Across a well-executed 12-week block, most runners move 1 to 2 VDOT points. At VDOT 46 that is around 4 to 8 seconds per kilometre at threshold, which is real but small, and it means any apparent change smaller than that is almost certainly weather, course or pacing rather than fitness.
Two signals are worth watching between races. The first is whether easy-pace runs are getting faster at the same heart rate, which is the cleanest evidence of aerobic development and does not require a test at all. The second is whether threshold sessions at the prescribed pace are getting easier, which usually precedes a race result confirming it.
This is also where continuous data earns its place. A watch logs every run, so the drift in pace at a given heart rate is already sitting in your history; the difficulty is that nobody goes looking for it. Connecting your training history to an AI coach through the Garmin MCP server means you can simply ask whether your easy pace at 140bpm has moved since August, and get an answer from your own files rather than an impression. The marathon heart rate guide covers the pace-versus-heart-rate relationship in more depth.
Putting it together: a VDOT protocol
- Race something. A 5K is enough and is the easiest distance to run maximally. Pick a flat course and a cool day.
- Enter the result in the VDOT and running zone calculator to get your value and the five paces.
- Cross-check against a second distance if you have one from the same month. If the two disagree by more than two points, use the longer race for endurance paces and the shorter one for interval and repetition paces.
- Write the five paces down somewhere you will see them before a session, not just in the plan. Most pace errors are recall errors.
- Spend 80% of weekly volume at E. The distribution matters more than the precision of any single pace.
- Run one threshold session a week. If you only keep one quality session, keep this one.
- Recalculate after your next race, not on a schedule, and expect a move of 1 to 2 points rather than five.
A race result is a single data point. VDOT's real contribution is turning that one point into a week of decisions you would otherwise be making by feel, at exactly the moments when feel is least reliable. Start a 7-day free trial if you would rather have those decisions made against your full training history than against one 22:40.
