Warm up vs stretching is presented as an argument about which is better, and it is not an argument at all. They are two activities with different mechanisms, different timescales and different reasons to exist. One prepares you for the next hour. The other buys you range over the next six weeks. Almost every confusion in this subject comes from expecting the second to do the job of the first.
This page separates them, gives the numbers on what happens when you use one as the other, and describes exactly what Limber calls a warm-up — including a state in the code that nobody can currently reach.
The short answer
- A warm-up raises tissue temperature, moves blood and rehearses the pattern. It works in minutes.
- A stretch buys range through repeated time at length. It works in weeks.
- Long static holds immediately before a maximal effort cost about 5 per cent of strength and power.
- The threshold is around 60 seconds per muscle group. Below it, trivial.
- Put the same holds inside a full dynamic warm-up and the impairment disappears.
- Limber has 18 warm-up-eligible moves out of 56, and every one of them is dynamic.
What a warm-up is for
A warm-up has a job that finishes when the session starts. It raises the temperature of the tissue you are about to use, gets blood into it, and rehearses the pattern of the movement so the first working set is not the first repetition of the day.
Everything about that is short-horizon. It is undone by an hour of sitting afterwards. It leaves nothing behind at the end of the week. Judged as an intervention, it is preparation and nothing else.
The version that does this best is dynamic: movement through the ranges the session will use, in the directions it will use them, without hanging at the end of any of them. Behm and Chaouachi reviewed the acute effects of both static and dynamic stretching in the European Journal of Applied Physiology in 2011 and found the pattern that has held since — static holds before performance carry an acute cost, dynamic work does not.
Grade: reasonably solid.
What a stretch is for
A stretch has a job measured in weeks. Bandy and Irion, in Physical Therapy in 1994, gave 57 people with limited hamstring range 15, 30 or 60 seconds of static stretching, five days a week, for six weeks: 30 seconds beat 15 and matched 60. Their 1997 follow-up, with 93 people, found one 30-second hold a day matched three. Thomas and colleagues pooled the chronic trials in the International Journal of Sports Medicine in 2018 and found total weekly stretching time was what tracked the gain.
Grade: solid, and narrow. Young healthy adults, mostly one muscle group, six-week windows, on the easiest case anyone could have picked.
What that gain consists of is worth knowing. Konrad and Tilp, in Clinical Biomechanics in 2014, ran six weeks of stretching, confirmed the range increase, then measured the muscle and the tendon and found neither had changed detectably. Grade: reasonably solid. What moves first is how much stretch your nervous system will permit, not the length of the tissue.
Nothing in that description is preparation. A tolerance built over six weeks is not something you top up in the four minutes before a squat session.
Warm up vs stretching, side by side
| Compared on | Warm-up | Stretch |
|---|---|---|
| Timescale | Minutes | Weeks |
| Mechanism | Temperature, blood, rehearsal | Tolerance at length |
| Before an effort | Helps | Small cost |
Three columns is the limit of an honest table, so the fourth row goes in prose: the two do not substitute for each other in either direction. A dynamic warm-up will not build range, because it never spends time at end range. A long hold will not prepare you, because what it leaves behind is a muscle producing slightly less force.
The practical rule that falls out is short. Move before. Hold after, or on a day of its own. Stretching before or after a workout works through the ordering, and static versus dynamic stretching covers the two techniques in general.
The decrement, and the rescue
The numbers, because the internet exaggerates them in both directions.
Simic, Sarabon and Markovic meta-analysed the acute effects in the Scandinavian Journal of Medicine & Science in Sports in 2013 and found static stretching immediately before a maximal effort reduced strength, power and explosive performance by around 5 per cent. Behm and colleagues, reviewing the field in Applied Physiology, Nutrition, and Metabolism in 2016, put the practical threshold at about 60 seconds of stretching per muscle group — below that, the effects are trivial.
Grade: solid.
Then the finding that settles the argument. Blazevich and colleagues, in Medicine & Science in Sports & Exercise in 2018, embedded static stretching inside a full dynamic warm-up and measured athletic performance afterwards. There was no impairment. The decrement in the earlier studies came from stretching and then walking straight to the test.
Grade: solid for the rescue.
So the honest instruction is not "never stretch before training". It is: if you hold something before training, do the warm-up after the holds, not before them. And note that almost nobody stretching casually crosses 60 seconds per muscle group anyway, which is why this finding sounds more alarming in print than it is in a gym.
One thing neither activity does is prevent injury. Lauersen, Bertelsen and Andersen pooled the randomised injury-prevention trials in the British Journal of Sports Medicine in 2014 and found strength training reduced sports injuries to under a third and overuse injuries to about half, while stretching showed no significant protective effect. Pope and colleagues had randomised 1,538 army recruits to stretch or not before exercise in 2000 and found no reduction in injury. Grade: reasonably solid. Does stretching prevent injury is the long version.
What the app calls a warm-up
Limber's definition is narrow and checkable.
A warm-up step is a move flagged warm-up-eligible and dynamic — one repetition per breath cycle, six to twelve of them, rather than a hold. Eighteen of the 56 moves qualify, and all eighteen are dynamic. The engine takes one on a short session, two when the budget reaches twenty minutes, and picks whichever covers the most of the areas the session is about to work.
| Area | Warm-up moves | Example |
|---|---|---|
| Calves and ankles | 4 | M011 Wall Ankle Dorsiflexion Rock |
| Thoracic | 3 | M001 Cat-Cow |
| Hips | 3 | M012 90/90 Hip Switch |
| Hamstrings | 3 | M027 Toe-Touch Progression Rocks |
| Shoulders | 3 | M036 Standing Shoulder CARs |
| Neck | 2 | M045 Chin Tucks |
Read the gaps as well as the rows. No warm-up-eligible move is filed to the hip flexors, the glutes or the wrists as its own area. The engine reaches the first two only through a move's secondary areas, and the wrists not at all. A session led by hip flexors is warmed up by something adjacent.
Two more rules, both fixed. The contract-relax protocol is barred from a warm-up under every setting. And a five-minute routine skips the warm-up entirely and runs all-gentle instead, on the honest grounds that five minutes cannot afford a preamble.
Every session then closes with a downshift, and the catalogue has only three moves eligible for it: Child's Pose, Supine Twist and Happy Baby. None of them is a shoulder or neck position, so a shoulder session still finishes lying on the floor with the hips or the lower back. That is a gap in the library rather than a design decision, and it is worth naming.
The state nobody can reach
There is a pre-workout mode drawn in the code. Forty-five minutes before a training hour, the day flips: the app would offer a short priming routine beside the main plan, tagged with its own reason code, and the main plan would take two warm-up moves rather than one.
Nothing in the build writes that training hour, so the window never opens for anybody. And the routine it would hand you is a five-minute one, which by the engine's own rule contains no warm-up block at all. Drawn, unreachable, and not yet right.
It is described here as a sketch rather than a feature, for the same reason the training-reactive machinery is: a gate with nothing behind it is still a promise with nothing behind it. Limber writes a finished session to Health and never reads a workout back, so it cannot tell that you are about to train, or that you have.
Where to start
Take a reading before you rearrange anything. The toe touch test reads the posterior chain in five bands in about a minute, and the Range Score composes what you have measured over four of five stations, leaving the fifth blank because the test that would feed it is not built.
Then move before, hold after, and stop treating the two as versions of the same thing. Rest day mobility covers the day where only one of them applies, and the cluster sits under mobility for training.
Questions
Is stretching a warm-up?
No, and treating it as one is the central mistake in this subject. A warm-up raises tissue temperature and rehearses the movement, and it works in minutes. A stretch builds tolerance at length, and it works over weeks. Substituting the second for the first leaves you slightly weaker and no better prepared.
Should I do a dynamic warm-up or static stretching before training?
Dynamic, through the ranges the session will use. Static holds immediately before a maximal effort cost around 5 per cent of strength and power, with the threshold at roughly 60 seconds per muscle group and trivial effects below it. If you want holds beforehand, put the dynamic work after them rather than before, which is exactly the arrangement that removed the impairment in a 2018 trial.
How long should a warm-up be?
Long enough to raise temperature and rehearse the pattern, which for most people is a few minutes rather than a routine. Limber uses one dynamic move on a short session and two when the budget reaches twenty minutes, chosen for how much of the session's own areas they cover. There is no trial establishing an optimal length, and anyone quoting one is estimating.
Does a warm-up prevent injury?
Not on the evidence for stretching, which is the part that has been tested. Pooled randomised trials found stretching had no significant protective effect while strength training cut injuries substantially, and a trial of 1,538 army recruits stretching before exercise found no reduction at all. A warm-up is worth doing because the first set feels better, which is a reason that does not need a trial.
Can I build range with a dynamic warm-up?
Not much. Range adaptation comes from repeated time spent at end range, and a dynamic move by definition passes through the range rather than staying in it. Limber uses its dynamic moves at the front of a session for exactly that reason, and puts the long holds in the main block where they can do the work.
Why does my session end with the same three positions?
Because only three moves in the 56-move catalogue are eligible for the closing downshift: Child's Pose, Supine Twist and Happy Baby. The engine always reserves one for last, so the ending repeats more than the middle does. It is a thin shelf and naming it is more useful than pretending the variety is there.