Does Pausing the Timer During Rests Kill Your Garmin Training Load?
Everyone says pausing the timer between sets deletes your Training Load, and I said it too. Then I checked it against the Firstbeat model, which scores a session on peak EPOC rather than by summing seconds, and which says both that longer duration raises EPOC and that shorter rests raise it. Pausing does both at once, so the two effects point opposite ways and nobody has measured which wins. Here is what the model actually says, the one case that is not ambiguous, and the paired session that gets you your own answer.
For about two years I paused the timer during every rest on strength days. It felt like the responsible thing. Only record real working time, keep the file clean, no dead minutes padding the session out.
Then I picked up the idea that this was quietly deleting most of what Garmin had to score, and I repeated it. On this site, and on Garmin’s own forums.
So I went and checked it against the model instead of against my intuition. The short version: the mechanism I was describing is not the one Firstbeat documents, and once you use the right one, the answer stops being obvious. It is still worth changing what you do. Just not for the reason I gave.
What Training Load is actually built from
Training Load comes from EPOC, the recovery cost the Firstbeat model estimates from your heart rate. Your seven day Training Load is the sum of the scores from your individual activities, so that part of the usual explanation is fine.
Inside a single activity it works differently than most of us say it does. Firstbeat’s white paper is blunt about it: to determine the training effect of a single exercise, the peak EPOC achieved during that exercise must be known. Two of the figure captions repeat it. It is the peak the model reaches, not a running total of every second in the file.
That is the bit I had wrong, and it is the bit nearly every version of this advice gets wrong, mine included. “You deleted seconds, so you deleted load” is an appealing sentence about a model that is not adding up seconds.
Pausing pulls in two directions at once
Hold the pause question against the model properly and you get two effects, not one, and they point opposite ways. The same summary box in that paper lists both, one line apart.
The first: the higher the intensity and the longer the duration of exercise, the more the EPOC accumulates. Pause every rest and your file gets dramatically shorter, often three or four times shorter on a strength session. That pulls the number down. This is the effect everybody cites, including me.
The second, immediately below it: the shorter the recovery periods during exercise, the higher the EPOC. Pausing does not only shorten the file. It removes the easy minutes and splices your working sets together into something that reads as far more continuous than the session really was. That pulls the number up. Firstbeat demonstrates this one directly, with two matched sixty minute workouts at the same intensity, where the interval version scores lower than the continuous version precisely because of the low intensity recovery periods sitting between the hard bouts.
One effect says your paused session should score lower. The other says it should score higher. Both are from the same page.
Why nobody can tell you which one wins
Because the published work does not test the case we actually care about. Every comparison in it keeps the recording running and varies what happens inside it. Rest that is recorded and rest that is missing from the file entirely are not the same input to the model, and I cannot find anyone who has measured the second one.
I cannot settle it from Rack’s data either. Garmin computes these scores on the watch, and we never see the intermediate values, only what Connect ends up showing you.
One more caveat that cuts against my own citations. That paper is from 2005, last revised in 2012. Garmin acquired Firstbeat in 2020, and current watches do more than the paper describes. EPOC is still the documented basis, but I am not going to hand you a fourteen year old document and call it a spec for your 2026 watch.
The one version that is not ambiguous
If you never start an activity at all, because the lifts went into a phone app or a notebook, Garmin scores exactly zero. No setting fixes that and no model subtlety applies to it. This is still the most common reason a lifter’s load chart sits flat, by a wide margin, and it is worth separating from the much narrower pause question.
Worth one check while you are in there: if Auto Pause is enabled on your strength profile, standing still between sets is exactly the condition it watches for. It will make the pausing decision for you without asking.
Run it on yourself, it takes two sessions
Since the model does not settle it, the only answer I trust is the one with your numbers in it.
- Take a session you have done recently and know the shape of.
- Record it the way you normally do, pauses and all. Note the Training Load it produced.
- Repeat it, same exercises, same weights, roughly the same rest lengths, and let the clock run the whole way through. Stop it when you walk out, not when you rack the last set.
- Compare the two load numbers, and the two Recovery Time values the next morning.
Same body, same work, only the recording changed. Whichever way it lands, you will know something about your watch that I could not tell you from the paper. If it lands somewhere surprising, I would genuinely like to hear it.
My own position, for what it is worth: I stopped pausing. Not because I can prove it raises the score, but because a continuous file is the one that matches how the model was described and validated, and because the recorded rest is real information about the session. A shorter file with the easy minutes cut out is a less accurate record of what you did, whatever number comes out the other end.
How this squares with the other thing I have written
I have argued that Garmin underrates strength training load, partly because long rests drag your session average heart rate down into an easy band. That still stands, and it is a separate claim from this one. The average and the peak are different quantities, and it is the peak that sets the score for a session. Lifting reads low on a heart rate model for reasons that have nothing to do with your pause button, and nothing in this post makes heart rate derived lifting load accurate in an absolute sense. A heavy low heart rate session will still price below a moderate run, because the model is measuring oxygen cost and your quads were sending a different bill.
What Rack does with the clock
Since I build a strength app for these watches, the fair thing is to say what it does rather than imply it solves the model. Rack starts one recording when the workout begins and stops it once, when you finish. Rests are inside the activity. There is no pause between sets.
The finished session lands in Garmin Connect as a standard FIT activity over the same partner channel a bike computer uses, with heart rate captured per set and your rest intervals, reps and weights preserved in the file. That gets you a complete record of the session. It does not turn Training Load into a strength number, and anyone telling you their app does that is describing something that is not in the model.
You can also do all of this with no app at all. Start the native strength activity, leave it running, stop it when you leave. The recording habit is the free part.
Rack is on the Connect IQ store, and the iPhone app works on its own if you would rather start there.
Ran the paired session and got a result either way? Send me a message. I’m the dev, and I read everything.