Is TSS a useful metric for mixed difficulty training?

TSS is essentially IF (via normalized power) multiplied by 100 per hour. It will measure non-linear increases in RPE as just a linear increase in points. For example, if you perform a 2 hour super easy (0.45 IF) ride with a friend, that might calculate out to more TSS than a brutally hard 60 minute interval session which is unlikely to reach 0.9 IF. That really breaks its usefulness as a proxy to estimate how much training stress you’ve put in your body on a given ride or in a given week.

This raises a question for me: how is TSS supposed to be used? It seems like Garmin’s Training Load is a far more useful metric, as it does a better job capturing how much i need to recover from a training session or a rolling period of training.

Just a few facts to put your comments in context:

  1. TSS = IF squared x 100 x duration (in hours). Point being, it increases NON-linearly with intensity. In fact, as shown in one of the last few slides here, TSS closely parallels perceived exertion (better than Banister’s TRIMP or Rusko’s EPOC scores, anyway):
  1. An IF of only 0.45 is extremely low…in fact, I’m not sure I personally could ride that easily without falling over.

  2. A challenging 1 h interval workout will have an IF of more than 0.9…occasionally, even over 1.05 (i.e., an NP buster).

How long have you actually been using a power meter?

ETA: Garmin’s “Training Load” is based on Rusko’s EPOC.


no it won’t. For example…Dans in the TR catalogue has an IF of 0.5 for 30 minutes…which equals a TSS of 13. So if you did Dans for 2hrs, you’d accumulate 52 total tss. A 1hr threshold workout with an IF of 0.9 will give you +/- 81 tss. You’d have to do over 3hrs worth of Dans to get the same TSS as one 1hr threshold workout, but we all know that would be silly because it’s not a high enough zone to induce significant adaptation.

I’d bet if you matched your current CTL by only doing 0.5IF rides you’d actually end up suitably tired after a few months because it would be a ton of volume. So in that regard I’d say it’s a decent measure of stress. Though again, that training zone is sub-optimal.


4 hrs at 0.45 IF is the same TSS as 1 hr at 0.9 IF.

0.9 IF I think is hard, but not brutally hard. If it is brutally hard, your FTP may be set too high.

I do think, however, that the TSS metric needs to be “calibrated” for a given individual through training experience. Eg - a 0.9 IF workout (81 TSS for 1 hr) for me - a sprint-oriented athlete - is probably harder to recover from than an endurance-oriented athlete, as I probably burn more glycogen.

Ah, okay, thanks. i did misunderstand the calculation then. I still find it to be less representative of my training stress than garmin’s training load, because it still underrates very high intensity relative to moderate intensity.

Try doing some supramaximal (as in, requiring >100 of VO2max) efforts and see what happens. Like all other HR-based metrics, Russo’s EPOC (which is what Garmin’s Training Load uses) gives you absolutely zero credit for the additional stress - it can’t, because HR is capped by maximum HR.