SYNTH-55 · Guide
Cadence
Let the rider choose it, expect it to fall on climbs and late in long rides, use low-cadence work only as a hard-session accent, and never promise efficiency from pedalling drills
The question it answers
What cadence to ride at; whether a low cadence is a fault; whether low-cadence "strength endurance" or 90+ rpm spinning is worth training; why cadence drops on climbs and late in long rides; whether to raise running cadence.
Default (what to prescribe first)
Ride at the freely chosen cadence on every kind of ride, and cue a cadence only when the session is built around one. This rests on three findings:
In 7 experienced cyclists, freely chosen cadence did not differ from the cadence that minimised blood lactate at the same power, both on the flat in the drops (89.3 vs 87.7 rpm) and uphill upright (82.1 vs 81.5 rpm). The authors say "most but not all" riders chose near their optimum (REF-262, B, treadmill study).
At a constant 2.5 W/kg, metabolic cost was lowest at 60 rpm but riders chose 81 rpm. The chosen cadence gave the highest vastus lateralis power capacity with no significant efficiency penalty (REF-238, B; sample size not in the abstract). The usual self-selected cadence sits above the most economical one, and the entry reads that as buying power reserve at little cost.
A systematic review of 13 studies concluded that no cadence range can be recommended for later neuromuscular performance. Its one consistent signal is that high cadence combined with high intensity impaired neuromuscular function and performance more than low cadence (REF-255, A-).
Not from a study; the base's own practice: an athlete asking "is my 75 rpm wrong?" gets "no, not by itself". Their cadence is theirs unless a specific reason in the table below applies.
Decision (who → what)
| Situation | Do this | Source and how sure |
|---|---|---|
| Easy / endurance ride | Freely chosen. A cadence well under 90 rpm is not a fault | In 14 professional track cyclists (Tier 4, lab ergometer, optima interpolated from 60/90/120 rpm), the optimal cadence was 66 rpm at LT1 and 76 rpm at FATmax (REF-258, B). The cadence with the least muscle activation was 57 rpm at 100 W and 70 rpm at 200 W (REF-237, B). Transfer from track pros and lab EMG to road amateurs is inference |
| Threshold and VO2max intervals | Freely chosen, "not unusually high". The optimum rises with intensity | REF-258 (B): 82 rpm at MLSS and 84 rpm at VO2max (track pros). REF-237 (B): least activation at 86 rpm at 300 W and 99 rpm at 400 W. REF-255 (A-): high cadence plus high intensity costs the most neuromuscular function. The 85-95 rpm cue in SYNTH-33 is house practice |
| Climbing | Let cadence drop several rpm and sit up. Do not ask for a flat-road cadence | Uphill upright, freely chosen cadence was about 7 rpm lower than on the flat and the optimal cadence fell with it (87.7 → 81.5 rpm), and maximal power was 2.0 ± 2.1% higher (REF-262, B, n=7, one gradient). In Grand Tours, 7 professionals averaged 71.0 rpm on ~15-km passes vs 92.4 in flat TTs and 89.3 on flat stages, at the same HR on climbs as in TTs (157 vs 158 bpm) (REF-884, B; what they chose, not a tested optimum) |
| Late in a long ride | Expect the comfortable cadence to fall. Do not correct it | After 55 min of cycling, the energetically optimal cadence fell (P < 0.01) in 14 experienced cyclists, whether they rode at >95 or <55 rpm (REF-261, B). The abstract gives no magnitude. Rides of 3-5 h are untested (inference that the shift continues) |
| Wants "strength endurance" / low-cadence work | Optional accent inside hard sessions only (about 50-70 rpm, or ~20% under preferred cadence); not moderate-intensity grinding as a main protocol, and not a strength or efficiency tool. Dose, who fits, indoor vs hill and knee red flags: SYNTH-82 | Review of 7 studies: no strong evidence of benefit, and the authors only "tentatively" suggest low-cadence bouts at moderate to maximal intensity (REF-117, A-). Small positive trials inside interval sessions (REF-084, 24 women; REF-259, 16 men; B+); positive uphill but matched by no-interval controls (REF-852, B); a null at 40 rpm and moderate intensity in veteran cyclists (REF-118, B+) |
| No gym, or wants on-bike strength | All-out efforts against very high resistance at very low cadence, 2×/week. This is resistance training, not tempo | In 37 well-trained cyclists this did not differ from full squats on any parameter over 10 weeks (REF-158, B+). In 46 well-trained men, all-out pedal cycles at 70% of maximal dynamic force improved Wingate power, MAP and thresholds. 4-min intervals at 35% did not change within-group (they beat control only on Wingate mean power), and the head-to-head was non-significant (REF-428, A-). Details: SYNTH-43 |
| Crit, track, cyclocross: needs fast legs | Some high-cadence intervals (~20% above preferred) | High-cadence intervals raised preferred cadence from 92 to 101 rpm and gross efficiency at 90 and 110 rpm (P = 0.05), but the low-cadence group gained more on the 15-min TT (REF-259, B+, n=16 men) |
| Indoor / ERG sessions and testing | Keep cadence consistent across tests and prescribe at the habitual cadence. Same watts at a different cadence gives a different threshold wattage, and (inference in REF-237, not tested) a different stimulus | Respiratory compensation point fell from 222 W at 60 rpm to 197 W at 100 rpm at a similar VO2 (REF-202, B, n=10 collegiate men). Muscle activation at fixed power changes with cadence (REF-237, B) |
| Triathlete: bike-to-run | Ride the cadence you race best at. A cadence trick will not rescue the run | In 9 well-trained triathletes, 20 min at 60, 80 or 100 rpm gave statistically indistinguishable 3-km runs. After 60 rpm, stride rate and speed in the first 500 m were lower (REF-656, B) |
| Masters | Same default. No age-specific cadence rule | The only masters cadence trial is the null for moderate 40-rpm work in 47-year-old veterans (REF-118, B+). The base holds no study of cadence choice by age (gap) |
| Novice (Tier 0-1) | Freely chosen. Coach gear choice rather than an rpm number | Less-trained non-cyclists (VO2max 44) chose lower cadences than cyclists at every power, falling from ~80 rpm at 75 W to ~65 rpm at 175 W, while cyclists and equally fit runners held 90-100 rpm; all chose above the most economical 53-60 rpm (REF-881, B, men, short lab bouts). Cycling experience did not decide it: fit runners pedalled like cyclists. Not from a study; the base's own practice: shift early and avoid grinding very low cadences on steep hills. No entry tests cadence coaching in novices (gap) |
| Runner: step rate | Self-selected by default. For knee or hip load sensitivity (e.g. PFP), cue +5-10% at the same easy pace and build it up over weeks (a transfer from healthy runners; no study in symptomatic runners) | In 45 healthy recreational runners, +5% and +10% step rate reduced energy absorbed at the knee, and +10% also at the hip. Injury and metabolic cost were not measured (REF-620, B). Uphill running raises step frequency and downhill lowers it (REF-546, B+). Across 10 studies a higher step rate consistently lowered vertical excursion, ground reaction force and joint energy absorption, with no injury outcome (REF-883, A-, mostly men). Experienced runners' preferred stride frequency matched the most economical one, fresh and after a hard hour (REF-885, B, n=16): the reason to change it is joint load, not economy |
How to progress / periodise
Low-cadence accent (optional). Place it inside a quality session the plan already holds, not as an extra session; the tested doses and the base's progression (practice) are in SYNTH-82.
High-cadence work is for events that need it (REF-259); placing it in the specific phase is the base's own practice.
Running step rate. Change it gradually, over weeks, with a metronome or watch. REF-620 is acute, so the retraining timeline is practice. Treat it as load redistribution, alongside sensible load progression. There is no evidence-based "10% rule" for weekly running increases (REF-565, B+, 4 studies, "very limited evidence").
When NOT to use it / red flags
Do not prescribe moderate-intensity 40 rpm "strength endurance" as a main protocol (50-60 rpm at moderate intensity is untested in the base; extending the rule to it is practice). In veteran cyclists it changed nothing (REF-118, B+); why it is not strength work and what else not to promise: SYNTH-82.
Do not stack high-cadence, high-intensity work before sprints or jumps the same day. Inference from REF-255's one robust finding.
Knee pain during low-cadence work → stop the accent. Not from a study; the base's own practice; the base holds no study of cadence and cycling knee pain. Check saddle height too: a review of 62 references found a 5% change shifts knee kinematics ~35% and moments 16%, a lower saddle seems to raise patellofemoral force, changes under 4% of trochanteric length seem not to matter, and the authors recommend 25-30° of knee flexion (REF-882, B+, mechanics, not injury counts).
Do not compare a threshold or ramp result done at an unusual cadence with the road (REF-202).
Do not prescribe "180 steps per minute". No entry supports a universal running cadence. REF-620 is a relative +5-10% change with no injury outcome.
What the evidence does and does not show
Shows (B, acute; experienced riders except [REF-237] and [REF-238], whose abstracts give no training status): the optimal cadence rises with intensity (REF-258, REF-237), falls uphill (REF-262) and falls after prolonged riding (REF-261). Most experienced riders self-select near it (REF-262), or somewhat above the metabolic optimum at little cost (REF-238).Pros dropped to ~71 rpm on long climbs (REF-884); less-fit men chose lower cadences than cyclists (REF-881).
Mixed (small training trials): positive for low cadence inside interval sessions (REF-084, women; REF-259, men), null for moderate 40-rpm work (REF-118), positive uphill but matched by controls (REF-852), heavy torque work not proven better than lighter work (REF-428); read trial by trial in SYNTH-82.
Does NOT show that pedalling-technique drills raise efficiency. Efficient riders spread power more evenly around the crank: dead-centre size correlated with gross efficiency at r = 0.75, and "force effectiveness" did not predict it (REF-739, B, cross-sectional, n=21).The entry itself warns against promising an efficiency gain from technique work, because the causal arrow is unknown. Indoor trainer riding changed muscle recruitment but not efficiency (18.5 vs 18.8%; REF-254, B+). The one cadence-training effect on efficiency in this base is cadence-specific: after high-cadence intervals, GE rose at 90 and 110 rpm (REF-259).Say "your spin at those cadences may get cheaper", not "you will become more efficient".
Evidence gaps
Novices and cadence coaching. REF-881 shows what less-trained men choose (lower, falling with power), not whether coaching an rpm helps them.
Masters and cadence choice. Only the moderate-intensity null in REF-118. There is no study of age and optimal or preferred cadence.
Field cadence in amateur racing. Only Grand Tour pros (REF-884); no amateur, power or gradient data.
Cycling cadence and knee pain. No entry; saddle height has a mechanics-only review (REF-882).
Size of the late-ride cadence shift and rides beyond 55 min (REF-261 gives no magnitude).
High-cadence drills (spin-ups, single-leg) as a programme. No entry beyond REF-259's intervals.
How this entry was checked and changed
Why this level
(acute lab studies at B/B+ on where the optimum sits; two A- systematic reviews on low cadence and neuromuscular fatigue; five small B/B+/A- training trials on cadence-specific intervals; nothing tests a cadence prescription for amateurs or novices)
Content correction 2026-09-27
Revision 2026-09-29: REF-881 (Marsh 1997, B: less-fit men choose lower cadences) sources the novice row; REF-884 (Lucía 2001, B: pros ~71 rpm on long climbs) the climbing row; REF-852 (Nimmerichter 2011, B, confounded) the low-cadence row; REF-883 (A-) and REF-885 (B) the runner row (change step rate for joint load, not economy); REF-882 (Bini 2011, B+, saddle height) the knee-pain flag. All agree with the default; gaps narrowed; level unchanged.
Verification 2026-10-01: every refId outside Cross-references and the revision log was opened in the KB.md snapshot (27 entries). All numbers checked against the entries match (rpm values, watts, percentages, n, levels B/B+/A-).
Verification 2026-10-01: no WRONG and no NUMBER findings. The weak points are over-absolute wording ('Never' moderate low cadence) and entry-labelled inferences stated as fact.
Verification 2026-10-01: Level B is acceptable: the Default rests on two acute B studies and one A- review, with the practice sentence labelled; the card itself says no trial tests a cadence prescription.
Verification 2026-10-01: CONTRADICTION with SYNTH-56 is recorded in SYNTH-56.json (SYNTH-56 lists 'low-cadence torque reps' among submaximal ERG sessions; this card allows low cadence only inside hard sessions or as all-out resistance efforts).
Verification 2026-10-01: REF-158 has no DOI of its own in the head block; the dose detail is from the companion MSS paper, as the entry states. Level line 'five small training trials': six are cited (084, 118, 259, 852, 158, 428) if on-bike resistance is counted; left unchanged.
Revision 2026-10-04: low-cadence detail moved to the new SYNTH-82 (low-cadence / torque riding).The 'strength endurance' decision row, the low-cadence progression bullet, the moderate-40-rpm red flag and the 'Mixed (small training trials)' evidence bullet are each cut to a short summary with a pointer; the softened wording from the 2026-10-01 audit is kept (REF-117 tentatively allows moderate-to-maximal intensity; no line says moderate low-cadence work does nothing).The on-bike resistance row, the knee-pain flag, the default, the title and the level are unchanged; every cited refId stays in Cross-references.