Training load for trail running: counting climbs and descents
To measure training load for trail running with lots of climbing, use two numbers: a training-stress score from grade-adjusted pace or heart rate, which counts each climb by the time and effort it costs, and a separate descent load, because those scores treat a steep descent as easy. Downhill running is metabolically less demanding than level running at the same speed, so pace and heart rate give it little weight, while your legs brake through every metre of it. This post compares the usual methods and works through two descents of the same 1,000 m that training stress and Pendio's Descent Load Index rank in opposite order.
TL;DR
- Training stress (TSS) counts a climb through the time it costs at your effort: in Pendio's grade-adjusted pace model a kilometre at +10 % is worth 1.55 flat kilometres, at +20 % 2.52.
- Scores built on pace or heart rate undercount steep descents: running at −20 % costs about half the energy of the flat (Minetti, 2002), while the leg muscles work eccentrically to brake.
- Two made-up descents of 1,000 m: at the same effort, 10 km at −10 % adds about 1.5 times the training stress of 5 km at −20 %, while the Descent Load Index gives the steeper one twice the load.
- Follow total load, climbing and descent as separate trends against your own usual weeks: a week in the mountains can leave total load flat while descent load jumps.
Why don’t distance and time capture a hilly week?
Distance and time say how long you were out, not what the ground asked of you. Weighted by how much each stretch’s grade slows you down, a flat 10K counts as 10.0 km of flat running and a 10 km out-and-back with 1,000 m (3,280 ft) of climbing as 20.1 km (pendio.app). Same 10 km on the map, twice the running in time.
So a trail load method has to count the vertical: the question is how, up and down.
What do the usual load scores count on a climb and a descent?
Each common method counts climbing in some way. None of the usual ones counts a steep descent for more than its time, its heart rate or how it felt:
| Method | How it’s worked out | A climb adds | A steep descent adds |
|---|---|---|---|
| Distance or time | Kilometres or hours | Time only | Time only |
| km-effort | Kilometres + D+ ÷ 100 | 1 km per 100 m up | Nothing |
| Session RPE | Your 0–10 rating × minutes | What you felt | What you felt |
| Heart-rate score (hrTSS) | Time and heart rate, against threshold heart rate | The heart rate it costs | Little: descending is metabolically cheap |
| Pace score (rTSS) | Time and graded pace, against threshold pace | The time it costs at your effort | The time it takes |
| Descent Load Index | Descent km × (grade ÷ 10 %)² | Nothing | More with the square of the grade |
The UTMB World Series defines km-effort as the distance plus 1 km for every 100 m of elevation gain (UTMB World Series). Session RPE, Foster’s method, multiplies a 0–10 rating of how hard the session felt by its minutes; a review of 36 studies found it valid and reliable (Haddad et al., 2017). Being one rating per session, it can’t say which part of a run was the descent.
How does training stress count climbing?
Training stress counts a climb through the extra time it costs at the same effort, so climbing is in the score even though the metres themselves aren’t added. In TrainingPeaks’ formula, TSS grows with the duration and with the square of the intensity factor (your effort as a share of threshold), and one hour at threshold scores 100 (Friel, TrainingPeaks).
For runs, rTSS takes the intensity from graded pace against your threshold pace, so “a hilly 40-minute run at a steady pace of 8:00/mile would generate a higher TSS than a flat version of the same run” (McGregor, TrainingPeaks).
Pendio’s rTSS uses its grade-adjusted pace (GAP), a fit to Strava’s heart-rate-based GAP model of 2017 (guide). At the same effort, a kilometre at +10 % takes 1.55× as long as on the flat and at +20 % 2.52× (guide), and the score grows with the time. Any GAP has two limits: it “does not account for the technical difficulty or condition of the terrain” (Strava Support), and it turns climbing into time, so a week’s D+ is still worth following on its own.
Why does a descent need its own number?
A steep descent is cheap for your heart and lungs and hard on your legs, and a score built on pace or heart rate sees only the first part. Minetti and colleagues measured the energy cost of running at 3.40 J/kg/m on the level and 1.73 at −20 %, about half (Minetti et al., 2002).
The muscles tell a different story. A 2020 review describes downhill running as “metabolically less demanding than level running and uphill running for a similar running speed”, and as “characterized by repeated and prolonged eccentric muscle actions of lower-limb muscles”: muscles working while they lengthen, to brake each step (Bontemps et al., 2020). It names slope, duration and speed as the main factors in how much muscle damage a downhill run causes, and finds that the people least used to downhill running generally had the most. CTS coaches say it plainly: GAP and Normalized Graded Pace “underestimate the overall stress of downhill running” (Rutberg and Koop, CTS).
The Descent Load Index (DLI) is Pendio’s answer: descent distance weighted by (grade ÷ 10 %)², so 1 km at −10 % is 1.0 and at −20 % it’s 4.0, in units of its own whose weekly trend is what matters (guide). The square is a design choice, not a figure from the research, which names slope as a main factor without giving a formula.
Two descents of 1,000 m: which one counts for more?
Of two descents that each lose 1,000 m (3,280 ft), training stress counts the long gentle one for more, and descent load the short steep one, by twice as much. Here are two made-up ones, run at an example effort worth 6:00 /km on the flat by a runner with an example threshold pace of 4:30 /km (an intensity factor of 0.75):
| Descent A | Descent B | |
|---|---|---|
| Length and grade | 10 km at −10 % | 5 km at −20 % |
| Height lost | ↓1,000 m | ↓1,000 m |
| Flat-equivalent distance | 8.7 km | 5.8 km |
| Time, and pace on the watch | 52:12, at 5:13 /km | 34:48, at 6:58 /km |
| Training stress it adds | about 49 | about 33 |
| Descent Load Index | 10.0 | 20.0 |
The flat kilometres come from the time factors in the guide: 0.87× at −10 %, 1.16× at −20 %. Training stress follows the time, so the gentle descent adds about 1.5 times as much. The descent load follows the steepness: on a steady grade, the DLI is the height lost in kilometres times the grade in per cent, 1 × 10 = 10 against 1 × 20 = 20. D− alone, 1,000 m both times, can’t tell them apart.
Neither number is wrong. Training stress measures how long and how hard the effort was; descent load measures how much steep braking your legs did. A week that holds one steady can hide a big change in the other.
How do you follow hilly load from week to week?
Follow each number as a trend against your own usual weeks. In TrainingPeaks’ Performance Management Chart, fitness (CTL) is “an exponentially weighted average of the last 42 days of training”, fatigue (ATL) the same over 7 days, and form (TSB) fitness minus fatigue (TrainingPeaks). Pendio’s iPhone app keeps the same three from training stress, Banister’s model as Coggan’s PMC (guide).
The acute:chronic workload ratio compares your last 7 days with your last 28. The app keeps one for total load, one for descent load and one for run climbing, because “descent can spike while the total stays calm” (guide). With invented round numbers: your usual week is 400 TSS with a descent load of 10, then a week in the mountains brings 400 TSS again and a descent load of 25. The total-load ratio stays at 1.0; the descent lane shows a week two and a half times your usual.
A ratio describes a change; it doesn’t forecast what the change will do. A 2020 critique found “no evidence supporting the use of ACWR in training-load-management systems or for training recommendations aimed at reducing injury risk” (Impellizzeri et al., 2020).
What should you compare your training with?
Compare your training with the race you’re training for: its climbing, its descending and how steep both are.
The free GPX course analyzer reads a race’s GPX in your browser: D+ and D−, the hardest climb and descent, D+ per 10 km and the terrain in 13 grade bands. Trail running calculators explains grade-adjusted pace and flat-equivalent distance in more depth; why your pace is slower on hills shows how to judge a hilly run by grade-adjusted pace; downhill running load counts descents by hand and on real courses. In the app, readiness sets your recent training against the course’s distance, climbing and terrain as a 0–100 heuristic that describes the gap and isn’t a forecast (guide). How to close the gap is for you or your coach to decide.
Sources
- Minetti AE et al. “Energy cost of walking and running at extreme uphill and downhill slopes.” Journal of Applied Physiology 93(3): 1039–1046, 2002. doi.org/10.1152/japplphysiol.01177.2001
- Bontemps B, Vercruyssen F, Gruet M, Louis J. “Downhill running: what are the effects and how can we adapt? A narrative review.” Sports Medicine 50(12): 2083–2110, 2020. doi.org/10.1007/s40279-020-01355-z
- Impellizzeri FM et al. “Acute:chronic workload ratio: conceptual issues and fundamental pitfalls.” International Journal of Sports Physiology and Performance 15(6): 907–913, 2020. doi.org/10.1123/ijspp.2019-0864
- Haddad M et al. “Session-RPE method for training load monitoring: validity, ecological usefulness, and influencing factors.” Frontiers in Neuroscience 11: 612, 2017. doi.org/10.3389/fnins.2017.00612
- Friel J. “Estimating Training Stress Score (TSS).” TrainingPeaks, 2009. trainingpeaks.com
- McGregor S. “Running Training Stress Score (rTSS) Explained.” TrainingPeaks, 2008. trainingpeaks.com
- TrainingPeaks. “What is the Performance Management Chart?” 2015. trainingpeaks.com
- Rutberg J, with Koop J. “Ultramarathon Runners’ Toolkit for Monitoring Training.” CTS, updated 2025. trainright.com
- Strava Support. “What is Grade-Adjusted Pace (GAP) on Strava?” Checked 19 October 2026. support.strava.com
- UTMB World Series. “About the UTMB World Series.” Checked 19 October 2026. utmb.world
- Pendio. “How Pendio works”: training on your phone and split times. pendio.app/guide
- Pendio. Home page: the flat 10K and the 1,000 m grind. pendio.app
Questions
Does TSS account for elevation gain?
Indirectly. rTSS comes from graded pace, so a climb raises it through the extra time it takes at your effort; a heart-rate score rises with the heart rate it costs. Neither counts a steep descent's load on the muscles beyond its time and heart rate.
What is a good weekly descent load?
There isn't a universal one. The Descent Load Index is in units of its own, so what matters is the trend, how this week compares with your usual weeks, and how both compare with the descents of the race you're training for.
Can descent load tell me whether I'll get injured?
No. It describes how much steep downhill you ran, not what it did to your body, and whether load ratios predict injury is disputed in the research. If a descent leaves you in pain, talk to a doctor or physio.
The bottom line
Count hilly training in at least two currencies. Let training stress from grade-adjusted pace or heart rate carry the climbing and the overall effort, keep descent as its own number weighted by steepness, and judge both against your own usual weeks rather than a fixed target. Then set them next to the race: drop its GPX in the free GPX course analyzer to see its D−, its hardest descent and how much of it is steep.