SYNTH-44 · Guide
Strength training for runners
Heavy, few reps, short of failure, twice a week for at least eight weeks, then kept at a lower dose in season; it buys economy and time-trial speed, not VO2max, and circuits are a fallback, not the default
The question it answers
"What should I do in the gym?" — the TYPE of strength work, its load and volume, weeks to effect, what happens in season, and what changes for masters, women and first-time lifters.
Where it sits relative to key runs is SYNTH-31; how much interference to expect is SYNTH-21; plyometrics are SYNTH-45; isometric holds (tendon stiffness, an optional heavy calf add-on for economy that is plausible but not proven, resting blood pressure) are SYNTH-88; slow no-relaxation sets at a light load (the static-dynamic method) are not a substitute for the heavy default — no trial tests them in trained endurance athletes, and 4 weeks of slow bodyweight sets gave soldier-runners no significant 3000 m gain over controls (REF-1181, B+) — see SYNTH-89.
Default (what to prescribe first)
For a healthy Tier 1-3 distance runner more than ~8 weeks from the A-race:
Heavy lower-body strength, 2 sessions a week, for at least 8 weeks. Across 28 controlled studies in running, cycling, cross-country skiing and swimming, added strength improved middle- and long-distance performance (SMD 0.52 [0.33-0.70]), energy cost of locomotion (0.65) and maximal force (0.99); maximal-force (heavy) training beat other intensities, and the performance benefit held at every athlete level (REF-074, A-).In 24 controlled studies of distance runners, ≥4 weeks of added strength (heavy, explosive or plyometric) generally improved economy by 2-8% — a spread across studies, not a pooled estimate, and not in every study — and tended to improve 1.5-10 km time trials; the authors recommend 2-3 sessions a week mixing modalities (REF-072, A-) — here the heavy sessions plus SYNTH-45's small plyometric element.Not from a study; the base's own practice: "at least 8 weeks, 2 a week" reads the ≥8-week practice note attached to REF-074; it is not a tested threshold.
Which kind, by speed and outcome: in controlled trials of middle- and long-distance runners (195 moderately trained, 272 well trained, 185 highly trained), heavy loads (≥80% 1RM) improved economy a small amount (ES −0.266, GRADE moderate) and combined methods a moderate amount (−0.426, low); plyometrics helped only at ≤12 km/h (−0.307); submaximal (40-79% 1RM) and isometric work did not significantly (isometric: three small calf-only studies, ES −0.269, CI crossing zero); two calf trials from one lab (90% MVC, 3-s contractions, 14 weeks) saw running cost fall about 4% (REF-1155, B; REF-1156, B, not significant against controls); the heavy effect grew with speed and VO2max (REF-801, A-; lab economy; non-randomised trials included).On time trials and time to exhaustion (324 moderately trained, 272 well-trained, 298 highly trained runners), high loads were moderately effective (ES −0.469), combined methods largely (−1.035), plyometrics alone not significant (−0.210, p = 0.064); no method changed VO2max or vVO2max, and no moderator — sex, level, weeks, sessions a week — mattered (REF-865, A-; certainty very low to moderate; same team as REF-801, likely shared trials: one body of evidence).A heavy-only runner RCT: 17 well-trained runners (9 men, 8 women), half-squats 4 × 4RM 3×/week for 8 weeks — economy +5.0%, time to exhaustion at MAS +21.3%, VO2max and body weight unchanged (REF-862, B).Explosive work replacing 32% of training for 9 weeks improved 5 km time and economy in 10 well-trained endurance athletes (REF-802, B; sport not stated; a large swap, not a dose).
Session shape — not from a study; the base's own practice: 3-4 exercises (squat or leg press, hip hinge, single-leg movement, heavy calf raise), 3-4 sets of 4-8 reps with about 2 reps in reserve, ~3 min between hard sets, 30-45 min.From entries, none in runners: with sets to failure, >60% 1RM loads raised 1RM more than light loads, hypertrophy similar (REF-677, A; at fibre level adding load gave no evidence of a change in overall fibre growth in 8 small biopsy trials of mostly untrained young adults, and a lean toward more type II growth with heavier loads had every 95% CI including zero, REF-1195, A-); stopping sets at ≤25% velocity loss beat >25% with 212 vs 384 reps (REF-516, A, 308 resistance-trained young men); a 20% cap matched 40% on 1RM with 40% fewer reps and a larger jump gain (9.5% vs 3.5%) (REF-584, B+, 22 young men); 3-min rests built more 1RM than 1-min rests (REF-679, B+, 21 trained men)."2 reps in reserve" is the base's translation of the velocity-loss cap.It matches the line a 2026 summary by the leading resistance-training groups draws for muscle growth — no more than 2 reps in reserve, true failure adding nothing, rests of at least 60 s and 2 min on multi-joint lifts (REF-1143, B-, second-hand from meta-analyses in mostly non-athletes); the ~3-min rest here is set for strength, not size.
Expect economy, not VO2max. VO2max, velocity at VO2max, blood lactate and body composition were typically unaffected (REF-072) — so say "the same pace will cost less", and do not promise added mass or a higher ceiling: in 20 competitive distance runners over 40 weeks (11 lifting, 9 controls), the lifters' strength, economy and vVO2max were up at weeks 20 and 40 with no change in body mass, fat or lean mass (REF-863, B; programme not in the abstract).The gain can show up late in a run rather than fresh: in 28 well-trained men with no leg strength work in the previous 6 months, 10 weeks of twice-weekly heavy squats and single-leg press (3 × 6-8 at 65-80% 1RM progressing to 3 × 4-5 at 85-90%), heavy isometric calf raises and two jump drills left fresh economy unchanged but cut the rise in oxygen cost over a 90-min run at ~80% VO2max from 4.7% to 2.1%, and a run to exhaustion at 95% vVO2max straight afterwards lasted 35% longer (247 → 324 s), while the running-only group did not improve (REF-1144, B+; completers only, the lifters trained 0.8 h a week more, an open-ended test rather than a race).
Decision (who → what)
| Situation | Prescription | Source and how sure |
|---|---|---|
| Healthy runner, gym access, ≥8 weeks to A-race | The default above | REF-074 A-, REF-072 A-: direction; dose = house practice |
| Triathlete, run economy the target | Same heavy default, concurrent with full endurance training | REF-082 B: 15 triathletes, 14 weeks, 2 heavy sessions/wk improved maximal strength and running economy vs endurance-only, VO2max unchanged, hopping power preserved (it fell in the endurance-only group); no economy magnitude in the abstract |
| In season, or <8 weeks to the A-race (the 8-week cut-over is house practice) | Drop to about 1 session every 7-10 days, keep the load heavy, cut sets | REF-149 B+ (dose from its full text); REF-800 B (well-trained cyclists, 6 vs 6, a transfer): one heavy session a week through the first 13 weeks of racing kept the strength and muscle-size gains of a 12-week build; REF-587 B+: in general populations strength and size held up to 32 weeks on 1 session/wk and 1 set per exercise *if relative load was kept*; athlete data insufficient |
| Tempted to stop lifting for the season | Don't — reduce | REF-744 B, elite cyclists, n=7/group (a cycling transfer): after 25 weeks of lifting, stopping for the first 8 in-season weeks lost squat jump, Wmax and Wingate power relative to controls (ES 0.49-0.84); REF-375 A: cessation lowered submaximal strength (SMD −0.62), maximal force (−0.46) and power (−0.20), more with longer breaks |
| First-time lifter | 2-4 weeks at moderate loads to learn the lifts, then the default; no heavy lower-body before a key run in the first weeks | Not from a study; the base's own practice. Reason: after an identical bout, 82% of fibres in the eccentric arm were disrupted in untrained vs 45% in strength-trained men (REF-702, B; arm-curl model, two separate studies) |
| Very little time | One hard set per main lift, 2-3×/week | REF-586 A-: a single set of 6-12 reps at ~70-85% 1RM, 2-3×/week, to (near) failure still raised 1RM in resistance-trained men over 8-12 weeks (the one exception to the short-of-failure rule below); REF-583 A: with volume equated, frequency did not change strength gains (mostly untrained samples); REF-1141 C+ (non-systematic overview, non-lifters): one session a week or one set per exercise raised strength over doing nothing (~20% in single-set trials), less than full programmes |
| Masters (50+) | Default load, plus 1-2 sets moved as fast as possible (the set count is practice); 2 sessions/wk also in season; progress tendon-heavy work slowly | REF-597 A+ (adults ≥50, mostly untrained): high-velocity training raised leg power about three times as much as slow training (Δ 0.62 vs 0.20); REF-587: older people may need 2 sessions/wk, 2-3 sets to hold size; REF-599 B–C: tendon adapts slower — "progress slowly" is the entry's inference; REF-864 B: in 16 master marathoners (mid-40s, preliminary) maximal-strength training raised economy at marathon pace 6.17%, general resistance training did not; REF-1139 A-: in 451 adults at retirement age a heavy year (~70-85% 1RM) left leg strength at baseline 4 years on while light band circuits (mostly unsupervised at home) left no measurable advantage, and leg power fell in all groups; REF-1140 A- (abstract): in adults 60+ moderate-to-high weekly volume gave the most strength |
| Women | Same programme, not a lighter version; do not schedule it by cycle phase | REF-532 A (adults >50): women gained more relative lower-body strength than men (g = −0.21), men more absolute; REF-571 B: no reliable phase effect; REF-729 A-: the pill did not blunt strength or size gains |
| Postmenopausal woman who wants to keep muscle | Strength will rise; for muscle, more weekly sets | REF-523 B+: 41 women aged 52±4, 10 weeks, 2×/wk, 6-8 sets per muscle — 1RM rose in all groups, muscle grew only in premenopausal women; "more volume" is the authors' inference |
| No gym, bodyweight only | Progressive bodyweight / circuit work as a fallback, and say the evidence for economy is for heavy loads | REF-074: heavy beat lighter strength intensities; REF-801 A-: in runners, submaximal-load (40-79% 1RM) and isometric training did not significantly improve economy (isometric: three small calf-only studies; heavy isometric calf work is a possible add-on, REF-1155, REF-1156); REF-864 B: nor did general resistance training in master marathoners; REF-676 A-: bodyweight Tabata matched running Tabata in 334 male cadets — an equivalence result in non-runners, with no non-training control group |
| Current Achilles tendinopathy | Heavy slow calf-raise loading is the rehab, not a reason to stop all lifting — with a clinician | REF-626 A-: in 58 patients with chronic mid-portion Achilles tendinopathy, heavy slow resistance and eccentric training gave equal results at 12 and 52 weeks; compliance 92% vs 78% |
| Wants the gym to cut injury risk | Strength work, not stretching; for hamstrings (strides, hills, track), a low-volume Nordic block kept ≥6 weeks | REF-861 A-: 25 RCTs, 26 610 participants, general sports — strength RR 0.315, overuse injuries RR 0.527, stretching no benefit (heterogeneous); REF-887 A-: 13 studies, low- and high-volume Nordics gave similar eccentric-strength gains (g = 2.28 vs 2.12) after ≥6 weeks — adaptations, not injuries; REF-766 A: programmes including the Nordic about halved hamstring injuries |
How to progress / periodise
Weeks to effect. The reviews included interventions of ≥4 weeks (REF-072) and ≥5 weeks (REF-071, A); runner primaries ran 8 (REF-862), 10 (REF-1144, fresh economy still unchanged at week 10), 14 (REF-082, triathletes) and 40 weeks, with gains at weeks 20 and 40 (REF-863).The effect sizes in REF-074 fall in the order force (0.99) → energy cost (0.65) → performance (0.52); the entry reads that as "strength moves first, race speed last" — do not judge a block by a time trial in week 3 (inference).
Tendons lag. Tendon stiffness rose with loading (SMD 0.70) in interventions of ≥8 weeks and depended on loading intensity, not contraction type; ≥12 weeks showed a higher, not significantly different, effect (REF-517, A, adults 18-50).A larger meta-analysis (61 studies, mostly untrained adults) found stiffness SMD 0.74 across all loading types (0.80 for resistance training), no difference between high and low relative intensity, and larger gains when training strained the tendon ~5% rather than ~3%, a contrast resting on three intervention groups (REF-1152, A-).Not from a study; the base's own practice: add plyometric or depth-jump volume only after a block of heavy loading (the one runner RCT of the combination started drop jumps in week 4 of concurrent, supervised heavy lifting, REF-1144) (SYNTH-45 owns that, including the acute plyometric warm-up of REF-297, B).
Progress load, not reps. Not from a study; the base's own practice: when all sets are done with 2+ reps to spare, add load; keep reps in the 4-8 band.
No progress after a block is not a verdict. Before calling a runner a low responder, check that the sessions were done, that the load actually rose, and sleep and food (REF-1143, B-: the checks the leading groups list for low muscle-growth response, before blaming biology; transfer to strength is inference); one before-and-after test cannot separate a low response from measurement noise (REF-441), so retest.If it is still flat, change one variable and name it — for example, switch from adding load to adding reps within the 4-8 band for one block: in one trial some people grew more muscle under rep-based progression and others under load-based (REF-1143, second-hand, hypertrophy outcome) — and judge it on the same test after the next block (the base's practice).
Base → in-season. 2/wk in base and early build; about 1 every 7-10 days once racing or in the last ~8 weeks (REF-149 full-text dose; the 8-week cut-over is house practice); keep a maintenance touch through the taper rather than zero (Not from a study; the base's own practice — SYNTH-34 covers the endurance taper and does not address lifting).
Placement: ≥24 h before a quality run (48 h after high-volume or eccentric work), ≥3 h from same-day running, and session order are SYNTH-31 §2. Not runner data: same-session pairing attenuated explosive-strength gains (REF-330, A, untrained-to-active adults) — Inference: lifting straight after a run trades away the power end.
When NOT to use it / red flags
Do not sell strength as proven injury prevention for runners. In general sports it cut injuries to about a third (REF-861, A-): the direction transfers, the size for runners is unknown.The base's running-injury review finds strong evidence for long weekly distance (in men) and a previous injury as risk factors (REF-564, A-); "protect the knee with strength" in that entry is the base's practice, not its finding. In the plyometric meta-analysis only one of 30 RCTs assessed injuries (REF-518, A).
Not a substitute for running volume — an economy lever added on top (the entry's reading of REF-072, whose trials added strength to running).
No heavy lower-body work in the 24-48 h before a key session or race, and no first-ever heavy block in the last weeks before an A-race (house practice; SYNTH-31).
Sharp joint pain, a tendon that worsens across sessions, or bony pain → stop that exercise and refer; bone and tendon management is outside this framework (REF-626 for Achilles rehab).
Do not take sets to failure as routine — not-to-failure beat higher-volume to-failure lifting for upper-body strength and 20-min row power in 43 trained rowers (REF-640, A-; volume and failure were confounded).
What the evidence does and does not show
Shows: added strength improves economy and 1.5-10 km time trials without lowering VO2max (REF-072, REF-071, REF-865); heavy beats lighter work (REF-074, REF-801); stopping costs strength and power (REF-375, REF-744); in one runner RCT economy held up better late in a 90-min run and fatigued time to exhaustion rose, with fresh economy unchanged (REF-1144, B+).
Does not show: heavy vs circuit training head to head in runners beyond one preliminary masters RCT of 16 (REF-864; REF-801 and REF-865 compare methods across trials, not within one; "heavy > other intensities" in REF-074 is a cross-sport subgroup), or maintenance-dose data in runners (the in-season rows transfer from cyclists, REF-744, REF-800, and general populations, REF-587).
Weaker than it looks: the 2-8% economy figure is a spread across small studies (PEDro 4-6) (REF-072), and the newest runner RCT found no fresh economy change after 10 weeks (REF-1144); the masters and women rows come mostly from non-athlete samples (REF-597, REF-532, REF-523) and likely overstate trained responses (REF-529, A: by the base's transfer rule, lower-Tier effects are probably overstated).
Evidence gaps
How this entry was checked and changed
Why this level
(direction rests on an A- running systematic review REF-072, an A- cross-sport meta-analysis REF-074, an A review REF-071 and two A- runner meta-analyses of strength type vs economy REF-801 and vs performance REF-865, plus small runner RCTs REF-862 REF-864 REF-1144; exercise choice, set/rep scheme and weekly layout are house practice, and the masters/women rows transfer from non-runner strength trials)
Content correction 2026-09-29
Revision 2026-09-29: folded in REF-801 (Llanos-Lagos 2024, A-: heavy and combined methods improve running economy, plyometrics only at ≤12 km/h, submaximal and isometric not), REF-802 (Paavolainen 1999, B: explosive-strength swap improved 5 km and economy) and REF-800 (Rønnestad 2010, B, cyclists: one session a week in season kept gains) — a 'which kind, by speed' Default line, the in-season and bodyweight rows, the evidence summary and level parenthetical; the Llanos-Lagos candidate gap is removed and the runner-trial and maintenance gaps narrowed; cross-references added.
Content correction 2026-09-27
Revision 2026-09-29: folded in REF-865 (Llanos-Lagos 2024, A-: high-load and combined strength improve runners' time trials, plyometrics alone not significant, no dose moderator), REF-862 (Støren 2008, B: 4×4RM half-squats 3×/wk for 8 wk, economy +5.0%), REF-863 (Beattie 2017, B: 40 weeks, no mass gain), REF-864 (Piacentini 2013, B: masters, maximal > general resistance), REF-861 (Lauersen 2013, A-: strength RR 0.315 vs stretching none, general sports) and REF-887 (Cuthbert 2019, A-: low-volume Nordic as effective as high) — Default, masters, bodyweight and new injury-prevention rows, red flag, gaps (the four candidate gaps closed or narrowed).Default and level unchanged; the injury red flag now says direction yes, size for runners unknown.
Verification 2026-10-01: all 41 cited REF entries read in the KB.md snapshot; every number attributed to REF-074, REF-072, REF-801, REF-865, REF-862, REF-863, REF-864, REF-802, REF-082, REF-516, REF-584, REF-375, REF-861, REF-887, REF-626, REF-517, REF-597, REF-532, REF-523 matches its entry; every level tag matches except REF-599 (B vs B–C).
Verification 2026-10-01: Level B+ accepted — direction rests on A-/A reviews and the parenthesis says the dose is house practice; of 16 Default/Decision rows 5 are direct evidence, 2 pure practice, 9 mixed (cyclist or non-athlete transfer, or evidence + practice dose). No high-severity finding.
Verification 2026-10-01: spacing numbers (≥3 h, ≥24 h, 48 h) agree with SYNTH-31; in-season dose agrees with SYNTH-43; plyometric speed pattern agrees with SYNTH-45.
Verification 2026-10-01: checked against the base's entries only; REF-801, REF-865, REF-861-864, REF-887 entries are themselves written from abstracts.
Revision 2026-10-15: folded REF-1139 (Bloch-Ibenfeldt 2024, A-, RCT n=451) and REF-1140 (Radaelli 2025, A-, abstract) into the masters row, and REF-1141 (Nuzzo 2024, C+, overview of minimal doses in non-lifters) into the very-little-time row; cross-references added. Defaults, decision rows and level unchanged.
Revision 2026-10-16: folded REF-1143 (Roberts 2026, B-, conference summary of resistance-training response heterogeneity, full text) into the session-shape line (its hypertrophy line of no more than 2 reps in reserve and at least 60 s / 2 min rests agrees with the default; ~3 min kept for strength) and a progression bullet on a flat block (check adherence, load progression, sleep and food, retest, change one named variable, e.g. rep- versus load-based progression for one block; the hypertrophy-to-strength transfer is inference); cross-reference added.Defaults, decision rows and level unchanged.
Revision 2026-10-16: folded REF-1144 (Zanini 2025, B+, RCT of 28 well-trained men without recent leg strength work, full text): the 'Expect economy' line now says the gain can show up late rather than fresh (drift over a 90-min run 4.7% → 2.1%, time to exhaustion afterwards +35%, fresh economy unchanged); added to the level parenthetical, 'Shows', 'Weaker than it looks' (no fresh change in this trial), the small-primaries gap and cross-references; REF-862 relabelled 'a heavy-only runner RCT' (no longer 'closest to the default'); the tendon line and weeks-to-effect line note that REF-1144 started drop jumps in week 4 of concurrent supervised lifting and still saw no fresh economy change at week 10.Defaults, decision rows and level unchanged.
Revision 2026-10-16: isometric economy lines reworded to "not significantly" with REF-1155/1156 as a possible add-on; tendon line extended with REF-1152.
Revision 2026-10-16: pointer to SYNTH-88 (isometric training) added to 'The question it answers' and Cross-references; the isometric economy and tendon lines themselves are reworded by the batch-B fold (iso/B/fold.json), not here. Defaults, decision rows and level unchanged.
Revision 2026-10-16: one sentence added to 'The question it answers' — slow no-relaxation (static-dynamic) sets are not a substitute for the heavy default; no trial in trained endurance athletes; 4 weeks of slow bodyweight sets gave no significant 3000 m gain over controls (REF-1181, B+) — with a pointer and cross-reference to SYNTH-89. Defaults, decision rows and level unchanged. (SYNTH-89)
Revision 2026-10-26: folded REF-1195 (Varovic 2026, A-, exploratory meta-regression of 8 small randomised biopsy trials in mostly untrained young adults, full text) into the session-shape line beside REF-677 and into Cross-References: adding load gave no evidence of a change in overall fibre growth, a type II lean with heavier loads was imprecise.Fibre area is not economy or strength, so the heavy default, which rests on economy and time-trial evidence, is unchanged; level unchanged (B+).