What the app does, and what the numbers can’t tell you.

Because weekly mileage is not the lever, and this is the single most useful thing on this page if you are training for a race.

What moves VO₂ max is time spent near your ceiling — above 90% of your maximum heart rate. A marathon build accumulates enormous volume and very little time in that zone. You can run forty miles a week for months and barely touch it.

When a finished workout comes back from Apple Health, we measure the time you spent above 90% and show it. If no workout has come back yet, or one came back without readable heart rate, the card says so rather than showing you a zero. What we will not do is tell you how many minutes is enough. We used to print a 12-to-16-minute range here; an adversarial review of our own rules found that figure had no citation behind it, and that the studies underneath it measured time above 90% of VO₂ max rather than of maximum heart rate — a different quantity. So the measurement stays and the pass mark is withdrawn until someone qualified sets one.

So more hours will not fix it, and Uthold will not pretend they will. Your weekly time budget decides how much easy running fits around the hard sessions; it does not change what we project you will gain. The hard sessions are what the projection rests on, and they are capped at two a week — three only if you have been training consistently for six months and your recovery checks have been in range — regardless of how much time you have. That cap is our own default rather than an established limit, and it is labelled that way in the rulebook.

No. The field test on this site runs in your browser with nothing but a stopwatch and somewhere flat to run or walk, and you can enter results by hand.

A watch helps, because time above 90% of your maximum heart rate is the thing worth measuring and a chest strap or watch is how you measure it. Without one you still get the plan and the sessions; you lose the per-session feedback on whether you actually hit the intensity.

Yes, and the app supports it. The catch is that changing activity changes your heart-rate targets.

Your maximum heart rate is not the same on a bike as it is running, and the gap is not a fixed number you can subtract. A heart-rate band derived from running can land at or above your actual maximum on a bike — which would turn a hard-but-doable interval into one nobody can complete. So we do not carry targets across modalities automatically, and we do not apply a universal offset, because there isn’t a defensible one.

Where a modality-specific target exists, we use it. Where one doesn’t, the session falls back to effort-based guidance within whatever safety restrictions apply to you. Effort guidance does not guarantee a particular percentage of maximum.

One session is the exception, and it is tied to a single machine: the two-effort all-out sprint session is stationary cycling only, because a stationary bike is what the study behind it used and an all-out effort is the worst place to improvise a substitute. Everything else travels.

$12 a month, or $89 a year — about $7.40 a month. That is the intended price when the iOS app opens.

There is no date. The app is being built, and saying a month we might miss would be the first promise we broke. The waitlist gets one email when it opens, and nothing before it.

Nothing is charged now and there is nothing to buy — the app has no account system and no checkout in it yet. The field test on this site is free and needs no signup.

No. Uthold is a general-wellness product. It does not diagnose, treat, cure or prevent any condition, and nothing in it is medical advice. If you have a reason to check with a doctor before training hard, check with a doctor.

What we do have, which a disclaimer is not, is a screening step that can actually stop you. Report chest pain or pressure, unusual breathlessness, dizziness, fainting or palpitations and the app stops all training — not just the hard sessions — until you confirm a clinician has cleared you. Other answers restrict you to light activity only, which is less than the moderate training an all-clear allows.

The same five symptoms appear inside every hard session. If you report one mid-workout the lock applies from that moment and covers everything — intervals, easy running and the field test alike — and only confirmed clearance lifts it. Not time passing, and not simply feeling better. It is a gate, not a notice you click past.

The field test on this site computes everything in your browser and transmits nothing. No account, no cookie from us, no record kept. Close the tab and it is gone. If you join the waitlist we store your email address and nothing else — your test result is never attached to it.

In the app, health data is read-only — the app never writes anything back to Health — and each type is asked for when the feature that needs it is first opened: VO₂ max, heart rate, resting heart rate and your date of birth at the start, then workouts, sleep and heart-rate variability only when you open the screen that uses them. Nothing is passed to advertising or analytics, because there is no advertising or analytics SDK in it. The privacy manifest declares zero collected data types.

We had written “never iCloud”, then found the stored health record was being swept into ordinary device backups anyway. We fixed it. The detail is in the privacy policy, still a draft pending legal review.

Probably, and we will tell you roughly how much as a range, never a single number and never a promise. Over twelve weeks at the recommended time budget, people starting below the 40th percentile for their age typically gain about 2 to 5 mL/kg/min. Between the 40th and 79th, about 2 to 4.5.

Above the 80th percentile the band we use starts at zero, deliberately. If you are already fit, a twelve-week block may not move the number at all, and an app that promises otherwise is guessing. Those bands describe what most people reach. They are not a floor, and individual responses vary a lot.

There is a second catch worth knowing up front: a field test cannot resolve a small change. See the next answer.

A field test estimates VO₂ max. It does not measure it. Only a lab test with a mask measures it.

The practical consequence is specific. For the Cooper 12-minute run, our estimated threshold for telling two comparable tests apart is about 320 metres, which works out to roughly 7.2 mL/kg/min. For the Rockport walk it is about 4.5. A retest that moves less than that, we treat as inside the noise and say so rather than congratulating you. That threshold is our own working estimate, not a settled figure. And note the reverse: a change we cannot detect is not the same as no change.

Watch estimates have their own problem — a wearable VO₂ max trend carries something like 13.3% error, which is wider than most of the gains anyone is trying to see. That is why the app leads on minutes above 90% of your maximum heart rate, which is measured directly, instead of on a noisy weekly VO₂ max line.

Yes, and most of the design work went into deciding what to leave off it.

It shows how your results have changed since your first test, and it does three things differently. Each source is kept as its own history — a Cooper result and a watch estimate measure different things, so they never share a line. Results are plotted as separate points rather than joined into a curve, because a smooth line through field tests draws confidence that isn’t in the data. And the test’s change threshold is drawn on the chart as a band, so you can see which of your results have actually moved enough to be distinguished from noise.

In practice a lot of them won’t have. That is the honest part: the post-race decline is usually visible on the chart and still inside the band. So the app tells you the drop is there and that it cannot yet be told from measurement noise, and then it stops — no “improving”, no fitted slope, no predicted date you’ll be back, no fitness-age curve. We draw the line where our own measurement stops being trustworthy and we don’t tell you a story inside it.

One practical note if you’re about to take your first test: treat it as a practice run. The Cooper result is dominated by pacing for anyone who hasn’t done one before, so an early “improvement” is often just you pacing it better the second time. Same test, same route, same time of day is what makes a comparison mean anything.

It is a reference-equivalent fitness age: the age at which the median person of your sex has the VO₂ max you just recorded. That is all it is. It is not your biological age, and it is not years added to or taken off your life — we will not put a number on either of those, because nothing here supports one.

The reference is the Norwegian HUNT study, which covers roughly ages 20 to 70 and up. Beyond those edges no exact number is honest, because the reference cannot place you. So the app shows nothing rather than handing you the nearest row, and the free test on this site shows a bound instead — “75+”, or “25 or under”. A bound is true of anyone past the table; a number would be a claim about you specifically.

That sounds pedantic until you see what the alternative does. Our own first implementation clamped to the edge of the table, so any male reading at or below 35.3 came back as “fitness age 75” however far below it went — meaning an 85-year-old with a genuinely poor result was told he had the fitness of a 75-year-old. Ten years younger, purely because the reference ran out. Our own software checks had been written to expect that result, which is how it survived so long.

It is worth knowing that other apps ship the clamped version as a feature. An app that tells you your fitness age is 23 and congratulates you on reversing ageing is usually telling you its table ran out. One more limit worth stating: HUNT's rows are age groups, not single years — the oldest is an open-ended 70 and over that cannot tell 72 from 88 — so a fitness age near the edges is coarser than a single number makes it look. Norms from Loe et al., PLOS ONE, used under CC BY.

Because that is what VO₂ max does, and it is why it has a sawtooth shape for anyone who trains in cycles. It climbs through a build, then falls back after the event. The fall is normal. It is also largely optional.

The useful finding here is about what you cut. In the research, cutting training volume by two thirds held VO₂ max for fifteen weeks. Cutting the intensity by a third lost it. So between events, one or two genuinely hard sessions a week can hold a number that took months to build, while everything around them gets much smaller.

Being straight about that evidence: it was measured in people who reduced their training, not in people who stopped and came back, and the subjects were young and previously untrained. It is a sound principle to plan around rather than a promise about your number, and how quickly fitness fades after fifty has not really been studied.

We ask, and if you do, we withhold heart-rate targets entirely rather than adjusting them.

Rate-limiting medication — beta blockers, verapamil, diltiazem, ivabradine and others — lowers your maximum heart rate by an amount nobody can predict from your age. There are published correction equations. The best of them, built for exactly this population, still carries around 18 beats per minute of error, which is wide enough to put a “90% of maximum” target somewhere unhelpful or somewhere unsafe. So we do not apply one.

Instead the app uses, in order: a current exercise prescription from your clinician if you have one, targets measured while on your current medication, or effort-and-talk-test guidance. If your medication or dose changes, stored targets are invalidated rather than carried forward.

Because we ship what we can defend, and a long protocol list is easier to market than to justify.

Tabata is the clearest example. The 1996 study used a cycle ergometer at about 170% of VO₂ max-equivalent demand, in seven young male physical-education students, four exhaustive days a week plus a fifth day of additional work. The speed-skating team people remember was the protocol’s coaching origin, not the study population. The twenty-seconds-on version marketed today inherits none of that and cannot inherit its results. When researchers tested the marketed cycling version properly — 65 relatively sedentary adults aged 18 to 28, eight weeks — fitness improved with no significant advantage over the comparison programmes, and lower enjoyment.

Our library is deliberately short, and every session in it carries an evidence grade you can read. Some are graded as unvalidated defaults, and we label them that way rather than dressing them up.