Beyond TSS: TRIMP, hrTSS and session RPE compared

Every training platform reduces your ride to a single number, and most riders never ask where that number came from. If you use power, it is almost certainly TSS. If you use heart rate, it might be TRIMP, or hrTSS, or a zone-weighted score. If you use neither, session RPE will do the same job for the cost of a five-second question.

These are not four ways of measuring the same thing. They are four different opinions about what makes a ride costly, and they disagree — sometimes by a factor of two on the same session. Knowing which way each one is biased is more useful than picking the "right" one, because the bias is systematic, which means it is predictable.

Here is what each method actually calculates, where it goes wrong, and which to reach for when.

What a training-load number is trying to do

The distinction that organises the whole field is between external and internal load. External load is what you did: watts, kilometres, metres climbed. Internal load is what it cost you: the physiological and psychological disturbance that actually drives adaptation. Two riders doing identical 250 W for an hour have the same external load and quite different internal loads, and it is the internal one that determines what happens next.

Power-based TSS is a hybrid — an external measurement scaled by your own threshold, which smuggles in a bit of individualisation. TRIMP and session RPE are attempts at internal load directly, one through the cardiovascular response and one through perception. None of them measures adaptation. They measure a stimulus and hope it is proportional.

The four methods

Banister TRIMP

The original. Training impulse takes the duration of the session and multiplies it by the fraction of heart-rate reserve you were working at, then multiplies again by an exponential weighting of that same fraction — with different exponents for men and women, derived from the blood lactate response. The exponential term is the clever part: it means an hour at 90% of reserve counts for far more than two hours at 45%, rather than the same.

Its weakness is that heart rate is its only input. Anything that moves heart rate without moving mechanical work — heat, dehydration, caffeine, poor sleep, the last hour of a long ride — inflates the score. Anything that produces enormous mechanical work without moving heart rate much, like a set of short sprints, barely registers.

Zone-weighted TRIMP

A simpler variant. Divide the ride into zones, give each zone a fixed weight, multiply minutes by weight, sum. Edwards used five zones weighted one to five. Lucía's version, developed on professional cyclists during grand tours, uses three zones bounded by the two ventilatory thresholds and weights them one, two and three.

The three-zone version is honest about something the others hide: it treats the two threshold crossings as the boundaries that matter, which maps onto how LT1 and LT2 actually divide the intensity range. The cost is coarseness. Every minute above the second threshold scores the same, whether you were at 105% or sprinting.

Power-based TSS

Take your normalised power for the ride, divide by your threshold power to get an intensity factor, and score the ride as duration multiplied by the square of that intensity, calibrated so that an hour at threshold is exactly 100. The squaring is the key design decision and it is what gives TSS its character: doubling the intensity quadruples the score.

Two things follow. Short hard sessions score relatively well, and long steady ones score relatively poorly — a four-hour endurance ride that ruins your Sunday can come in under 200. And because normalised power rewards variability, a ragged group ride scores higher than a smooth solo ride at the same average power. Whether that is a feature or a flaw depends on whether you think variability is costly. Often it is.

Session RPE

Ask yourself, about half an hour after finishing, how hard the whole session was on a 0–10 scale. Multiply by the duration in minutes. That is the entire method, it was published in 2001, and it has survived twenty-five years of people trying to beat it with better technology.

It has been validated against heart-rate-based methods across a range of sports and it keeps working. It needs no equipment, it captures things no sensor sees — how you slept, whether the session was mentally grim, whether you were riding into a headwind alone — and it costs nothing. Its weakness is that it is a single subjective number, so it drifts with mood and with how honest you are being with yourself.

Side by side

Where they disagree

Over a whole week, all four correlate strongly with each other, because duration dominates every one of them and duration is the same number however you measure it. The disagreements are per session, and they are largest at the two extremes: the very long easy ride and the short hard one.

Look at the middle group. On a one-hour session containing 25 minutes of VO2max work, Banister TRIMP scores it as little more than half a steady two-hour ride, because average heart rate never gets high enough for long enough. Session RPE scores it at one and a half times that ride, because it felt awful. Both are describing something true. Only one of them will stop you doing another one tomorrow.

What goes wrong in practice

Which to use

• If you have power, use TSS as the primary and session RPE as the cross-check. That combination catches the two failure modes that matter: TSS missing how you felt, and RPE missing what you did. When they diverge for more than a few days running, something is going on that neither alone would have shown you.
• If you only have heart rate, use a zone-weighted score rather than hrTSS for interval days. Both have the same lag problem, but the zone method at least does not pretend to be a power number. Our guide to training with heart rate only covers the rest of that setup.
• If you have nothing, session RPE is genuinely sufficient. When modelled against actual training responses, subjective methods have held up against objective ones more often than the technology industry would like.
• Do not mix them within one training block. Switching from TSS to TRIMP halfway through changes every weekly total and every ramp rate at once, and you will spend a month unable to tell whether your training changed or your arithmetic did.

The thing that matters more than the metric

Studies comparing these methods in competitive road cyclists have generally found that several of them relate sensibly to changes in fitness and performance, and that no one method is decisively better than the others. That is the least exciting and most useful finding in the field: the choice of metric matters far less than using one consistently and long enough to have a baseline.

What all of them share is a blind spot for the shape of the training. A weekly total tells you how much, not what kind, and two riders with identical weekly numbers can be building completely different engines. That is why the total is best read alongside the trend rather than on its own — Fitness, Fatigue and Form is one way to do that, and how many TSS per week you actually need is the companion on what the totals should look like.

Pick one. Write it down every day. In three months the number will start telling you things about yourself that no amount of choosing the theoretically superior formula would have.