Mountain Bike Repeated Climb Power Test: Scoring XCO Fatigue Resistance

A rider sets a personal best up the local four-minute technical climb on a Tuesday evening, fresh legs, clean effort, a number that would sit near the front of the category. Come Saturday's race, that same climb repeats five times over the lap course, and by lap three the rider is getting passed on the exact pitch where they were fastest three days earlier. Nothing about that Tuesday number predicted it. What separates the rider who still holds position on lap five from the one who gets shelled isn't how hard they can climb once. It's how much of that power survives four or five repetitions, each one launched after only a short, technical descent rather than a full recovery.

Most mountain bike testing measures the wrong half of that equation. An FTP test or a single hillclimb time trial captures peak sustainable output from fresh legs, which tracks fitness reasonably well but says nothing about what happens once that output has to repeat under the incomplete recovery an actual lap course imposes. The protocol below treats the climb the way a race actually presents it, repeated under partial recovery, and scores a rider on how much power survives from the first rep to the last rather than on the single best number buried inside the set.

Road climbing and cross-country climbing look similar from a distance, but they load the body differently in a way that matters for testing. A road climb is usually one continuous effort, paced from bottom to top, done once. An XCO lap is a loop, and the defining climb repeats on every single lap of the race, five to seven times in a standard format, with only a technical descent standing in for recovery between efforts. That descent is not passive rest either. It carries its own demand: braking, cornering, absorbing impacts, all while heart rate is still elevated from the climb that just ended.

A rider with a strong single-effort climbing engine can look identical to a well-rounded climber on a hillclimb time trial, then fall apart by the fourth ascent, because a one-off test never asks the legs to fire again before phosphocreatine stores and buffering capacity have finished recovering. Testing peak climbing power alone is a bit like judging a runner by their fastest single 400m repeat and never checking what the sixth one looks like.

Impellizzeri, Sassi, Rodriguez-Alonso, Mognoni, and Marcora (2002) monitored eight elite off-road racers during an actual cross-country event and found average heart rate sitting around 90 to 92 percent of each rider's laboratory-determined maximum across the full race, with the large majority of race time spent above individual lactate and ventilatory threshold. That is not one hard effort followed by recovery. It is a rider parked near their physiological ceiling lap after lap, exactly the kind of repeated near-maximal loading a single climb test can never reproduce. The study's own limitation is worth flagging: heart rate was the primary outcome rather than direct power output, since crank-based power meters were far less common in off-road racing in 2002, and the sample was eight elite men racing one course design. The precise shape of a repeated-climb power curve on a different course profile can't be read directly off that data.

Later power-meter-based race analyses, including Stapelfeldt, Schwirtz, Schumacher, and Hillebrecht's (2004) work on elite German racers, showed why the heart-rate picture looks the way it does. Race power output is not a steady-state line. It swings from near-recovery wattage on descents to repeated surges well above threshold on every technical rise and on every lap's defining climb, dozens of times across a single race. A rider's average lap power tells a coach almost nothing about whether the fifth climb looked anything like the first one.

The strongest published link between climbing ability and race outcome comes from Prins, Terblanche, and Myburgh (2007), who tested competitive cross-country riders in the lab and against a 6km uphill time trial, then compared both to actual race performance. Uphill time-trial time turned out to be one of the strongest predictors of race result among everything they measured, tracking performance more tightly than most of the laboratory variables tested alongside it. A rider's climbing legs matter more than almost anything else on race day.

But look closely at what that test asked the rider to do: one uphill effort, done once, from fresh legs. It identifies who has the best engine on the first climb, not who still has that engine on the fourth one, which is precisely the scenario that decides most XCO races once the lead group has thinned out. The sample was also a specific field of well-trained but not exclusively elite riders, so the correlation strength shouldn't be treated as a universal constant across every level of competition. That gap between fresh-legs climbing ability and repeated-effort durability is the entire reason this protocol exists.

• Power meter: A crank, pedal, or hub-based unit logging at least once per second. Use the same bike and position for every retest; even a different saddle height changes the power-to-effort relationship enough to muddy a comparison.
• Climb selection: A real outdoor climb 3 to 6 minutes long, ideally matching the gradient and surface of the target race, roughly 5 to 10 percent average grade. Mark exact start and end points and use the identical segment every session.
• Indoor alternative: A smart trainer with gradient simulation set to a fixed percentage works for controlled, weather-independent testing, though it loses the technical recovery-while-descending demand a real course adds.
• Heart rate monitor: Optional but useful as a secondary signal; a rider whose power falls while heart rate stays pinned near max is fatiguing differently than one whose heart rate is also dropping.
• Weather log: Heat above roughly 25 degrees Celsius measurably raises heart rate and perceived effort at a given power output, which can depress later reps independent of any real fatigue change.

1. Warm-up, 15 minutes: Ten minutes of easy spinning followed by three progressive one-minute efforts at roughly 70, 85, and 95 percent effort with two minutes of easy spinning between each, finishing with three to five minutes of easy riding before the first rep.
2. Climb rep one: Ride the marked climb at maximal sustainable effort, meaning race-pace intensity a rider could hold for that duration in competition, not an all-out sprint that blows up in the first 90 seconds. Record average power, average watts per kilogram, time, and average heart rate for the full climb.
3. Recovery: Descend or spin back to the start over a fixed time window that mirrors the non-climbing portion of the target race's lap, typically 3 to 4 minutes for a standard XCO course. Keep the effort easy and consistent, not a full stationary stop, since a real lap never gives a rider that.
4. Repeat: Complete five total climb reps for a standard XCO test (see the format table below for shorter or longer race scaling), recording every rep's power, time, and heart rate individually rather than only the first and last.
5. Valid trial criteria: Rerun the session if rep one is badly under-paced from an incomplete warm-up, if another rider or trail traffic forces a stop mid-climb, or if the recovery window runs noticeably long or short on any rep.

Total session time runs 35 to 45 minutes including warm-up for the standard five-rep version. Write each rep's numbers down immediately after finishing it; reconstructing five climbs' worth of power data from memory after the session is a reliable way to quietly corrupt the comparison.

The Power Maintenance Ratio, or PMR, compares average output on the final two climbs against the first two, using watts per kilogram so the score stays comparable across riders of different body mass: PMR% = (average W/kg, reps 4 and 5 ÷ average W/kg, reps 1 and 2) × 100.

Worked example: a rider posts 4.8, 4.7, 4.5, 4.2, and 3.9 W/kg across the five climbs. The first two reps average 4.75 W/kg; the final two average 4.05 W/kg. PMR% = (4.05 ÷ 4.75) × 100, which comes out to roughly 85 percent. That rider is holding a little over four-fifths of their fresh-leg climbing power by the last ascent, a moderate but not severe fade.

These bands are a field reference drawn from applied repeated-effort testing patterns rather than a single published normative study specific to this exact protocol, since the research above measures race demands and single-effort climbing separately rather than a five-rep power maintenance score. Weigh a rider's number against their own retest history on the identical climb before comparing it to anyone else's.

The standard five-rep, 3 to 4 minute recovery version fits a typical Olympic-format XCO race. Shorter or longer race formats call for different scaling so the test actually mirrors what the rider's legs will face.

The goal is matching the recovery window a rider's target event actually provides, not defaulting to whatever recovery feels comfortable. A marathon racer tested with a two-minute recovery is being scored on a demand profile that doesn't resemble their race.

• Changing the climb segment between sessions: A slightly longer or steeper substitute produces a PMR that reflects the course change, not fitness. Use the exact same marked segment every time.
• Letting recovery drift longer when a rider looks tired: Extending the rest window to let someone recover fully defeats the point of the test and inflates PMR artificially.
• Sandbagging the first rep: Pacing conservatively on climb one to save legs for later produces a PMR that looks better than real fatigue resistance, since the baseline was never a true maximal effort.
• Testing on inconsistent gradient or surface: Loose gravel one session and hardpack the next changes the power-to-speed relationship enough to confound a retest.
• Ignoring recent training load: A rider tested the day after a hard race or long ride will show a depressed PMR from accumulated fatigue rather than a genuine repeated-climbing limitation. Log training load from the prior 48 hours alongside every test.

A low PMR is a training target, not a verdict on a rider's engine. Riders in the moderate or poor bands typically respond well to structured repeated-climbing intervals rather than more steady-state Z2 riding: sets of four to six climbing efforts at 90 to 100 percent of climbing FTP, each lasting the same 3 to 6 minutes as the test climb, with recovery matched to the rider's target format from the table above. Two sessions a week for four to six weeks, layered onto normal aerobic volume rather than replacing it, is a reasonable starting dose.

If a rider tests well on a single hillclimb time trial but posts a poor PMR, resist filing that under a general fitness gap. It is a specific, trainable quality tied to how quickly phosphocreatine stores and buffering capacity recover under partial rest, not to peak aerobic power, and it is one of the few field measures that maps directly onto what happens once a race has repeated its defining climb three or four times and the group starts to string out.