Strength training improves running economy durability and fatigued high-intensity performance in well-trained male runners
10 weeks of twice-weekly heavy lifting plus jumps cut the economy drift over a 90-min run (4.7% to 2.1% from minute 15 to 90) and lengthened a run to exhaustion straight after it by 35%, while fresh economy did not change
Zanini et al. asked whether adding gym work to running makes economy hold up better late in a long run, and whether it helps a hard effort at the end.
Twenty-eight well-trained male runners were pair-matched on 10 km time and randomised to keep running as usual (E, n = 14) or to add two supervised 45-min sessions a week of plyometrics and heavy lower-limb strength work for 10 weeks (E+S, n = 14).
Before and after, they ran 90 min on a treadmill at 10% of the way from the first to the second lactate threshold (13.1 km/h, 79.7% VO2max) and then, without a break, ran to exhaustion at 95% of the speed at VO2max. The pre-to-post change in oxygen cost at minute 90 differed between groups: −2.1 ± 3.5% in E+S vs +0.6 ± 3.8% in E (p = 0.04).
Within E+S, the rise in oxygen cost from minute 15 to minute 90 fell from 4.7 ± 1.7% to 2.1 ± 1.0%; in E it did not change. Time to exhaustion rose 35 ± 25% in E+S (247 → 324 s) and did not change in E (194 → 169 s; −8% in the abstract), a group × training interaction of p = 0.004. Economy measured fresh (at 10 and 12 km/h and at minute 15) did not change in either group.
Leg-press 1RM rose 22%. Our reading: a real but small-sample signal that heavy lifting helps economy late in a run more than fresh; the time-to-exhaustion gain is large because that test magnifies small changes, and the control group got slightly worse.
Protocol
Participants: 38 enrolled, 28 completed (14 per group). Eligibility: men 18-45 y, 10 km under 50 min in the previous 6 months, more than 20 km a week, racing ≥5 km, used to running more than 90 min continuously at least 3 times a month, no lower-limb strength or power training in the previous 6 months, uninjured.Age 32.6 ± 8.4 y (E+S) vs 37.1 ± 8.7 y (E), body mass 71.3 vs 69.3 kg. Baseline 10 km speed 15.3 vs 15.4 km/h, VO2max 58.9 ± 7.5 vs 56.8 ± 5.7 mL/kg/min (the abstract gives 58.6 for the whole sample), fresh economy 215 vs 212 mL/kg/km; no baseline differences (p > 0.05).
Dropouts: illness 2, voluntary withdrawal 1, no response after the training period 3, injury caused by the training 1, injury unrelated to it 4. The split by group is not reported, and only completers were analysed.
Allocation: after baseline tests, pairs matched on 10 km best were randomly assigned to E or E+S. The method of randomisation is not described. No blinding of participants or testers is reported (participants cannot be blinded to lifting).
Gym programme (E+S), as reported (Table 1): 2 sessions a week, 2-4 days apart, ~45 min, every session supervised by a qualified strength coach. Standardised warm-up, then 2 plyometric and 3 lower-limb strength exercises, progressed every 3-4 weeks.
Vertical plyometrics: weeks 1-3 pogo jumps 3 × 10-12, 90 s rest; weeks 4-7 drop jumps 3 × 6, 120 s; weeks 8-10 drop jumps 3 × 6, 120 s. Ground contact under 200 ms while maximising height, with Optojump feedback after every rep.
Horizontal plyometrics (outdoors): weeks 1-3 hop and stick 3 × 6, 90 s; weeks 4-7 stiff-leg bounds 3 × 10, 90 s; weeks 8-10 bounds for length 3 × 8-12, 90 s.
Back squat and single-leg press (each): weeks 1-3 3 × 6-8 at 65-80% 1RM, 120 s rest; weeks 4-7 3 × 5-6 at 80-85%, 120-150 s; weeks 8-10 3 × 4-5 at 85-90%, 150 s. Squats lifted as fast as possible, load adjusted set by set to keep mean concentric bar speed at 0.35-0.55 m/s (which the authors equate with 80-90% 1RM); leg-press 1RM retested at the start of week 6.
Seated isometric calf raises: weeks 1-3 4 × 6-8 at 80-100% of maximal voluntary force, 60 s rest; weeks 4-7 and 8-10 5 × 4-6 at 100%, 90 s.
Adherence 99% (2 runners each missed 1 of 20 sessions).
Endurance training (both groups): each runner's habitual running, not prescribed or standardised by the researchers, self-reported in a daily written log with at least one GPS spot check per runner. Running did not change differently between groups (E+S 3.5 → 3.2 h and 47 → 43 km a week; E 3.7 → 3.5 h and 51 → 48 km a week; interaction p = 0.76).Total training rose 22% in E+S (3.9 → 4.8 h a week, +0.8 ± 1.1 h) and not in E (−0.1 ± 1.5 h); the total-volume interaction was not significant (p = 0.07). Endurance training was therefore similar in volume but not matched, and intensity was not reported. Self-reported strength training (type not stated) was 0.4 → 1.6 h a week in E+S and 0.2 → 0.3 h a week in E (Table 3).
Durability test (before and 3-7 days after the 10 weeks, same absolute speeds, same time of day ± 1 h): high-carbohydrate meal up to 2 h before, then water only; no caffeine, alcohol or intense exercise for 24 h; same shoes both times, carbon-plated shoes not allowed; same treadmill (1% incline stated for the incremental test, not restated for the 90-min run), 18-21 °C, 45-55% humidity, two fans.After 5 min at 10 km/h and 3 min at 12 km/h (fresh economy), 90 min at 10% Δ between LT1 and LT2 — 13.1 ± 1.4 km/h, 79.7% VO2max (the authors' "heavy intensity domain", described as marathon-like).Gas collected for 5 min every 15 min; economy = oxygen cost (mL/kg/km) from the last 2 min of each 15-min block, body mass interpolated across the run; water ad libitum after each measurement. Lactate, heart rate and RPE every 15 min.
Time to exhaustion: straight after minute 90, speed raised to 95% of the speed at VO2max from the first visit (16.1 ± 1.6 km/h), run to voluntary exhaustion with time and distance hidden.
Other tests: countermovement jump, unilateral leg-press 1RM, isometric plantar-flexor force, DXA body composition, before and after.
Analysis: three-way mixed ANOVA (group × training × run time) for economy, lactate, heart rate, RER and RPE; follow-up two-way ANOVAs on percentage change with Bonferroni post hoc tests; within-group drift relative to minute 15; two-way ANOVA for time to exhaustion.
Funding: National Strength and Conditioning Association Young Investigator Research Grant.
Key Results
| Outcome | E+S (n = 14) | E (n = 14) | Statistics (full text) |
|---|---|---|---|
| Pre-to-post change in oxygen cost at minute 90 | −2.1 ± 3.5% (226 ± 10 → 221 ± 14 mL/kg/km) | +0.6 ± 3.8% (218 ± 15 → 219 ± 14) | group × training × run time F = 3.81, p = 0.003, ηp² = 0.13; group × run time on % change F = 3.45, p = 0.006; between groups at 90 min p = 0.04 |
| Drift from minute 15 to minute 90, pre vs post | 4.7 ± 1.7% → 2.1 ± 1.0% (p < 0.001); at 75 min 3.4 ± 1.0% → 1.5 ± 1.9% (p = 0.003) | no training × run time interaction (p = 0.42); from Table 4 means ~2.8% → ~4.3% (our calculation) | within-group tests; E+S training × run time F = 6.14, p < 0.001, ηp² = 0.32 |
| Fresh economy at 10 and 12 km/h | no change | no change | interaction p = 0.86 and p = 0.89 |
| Economy at minute 15 of the run | 215 ± 11 → 216 ± 13 | 212 ± 13 → 210 ± 10 | interaction F = 1.03, p = 0.32 |
| Time to exhaustion at 95% vVO2max after the 90 min | +35 ± 25% (247 ± 94 → 324 ± 105 s, p < 0.001) | no change (194 ± 96 → 169 ± 106 s, p = 0.40; −8% in the abstract) | group × training F = 9.92, p = 0.004, ηp² = 0.28 |
| VO2 reached in the time to exhaustion | no change (p = 0.85) | no change (p = 0.96) | interaction p = 0.97 |
| Lactate at exhaustion | 5.6 ± 2.0 → 6.4 ± 2.4 mmol/L (p = 0.04) | 4.9 ± 1.8 → 5.3 ± 2.0 (p = 0.39) | interaction p = 0.38 |
| Lactate during the run | at 90 min 2.3 ± 0.4 → 2.2 ± 0.5 | at 90 min 2.4 ± 0.5 → 2.9 ± 0.8 | three-way F = 3.58, p = 0.02; E's pre-to-post rise larger at 75 and 90 min (p < 0.05); within E, post > pre at 75 and 90 min |
| RPE (6-20) at 90 min | 15.7 ± 2.0 → 14.4 ± 2.1 | 16.5 ± 1.5 → 17.3 ± 1.6 | three-way F = 3.62, p = 0.03; E+S larger drop at 90 min (p = 0.03) |
| Heart rate, RER during the run | no difference | no difference | three-way p = 0.83 and p = 0.60 |
| Energy cost (supplement) at 90 min | −1.8 ± 3.5% | +0.8 ± 3.9% | between groups p = 0.03; E+S drift +3.8 ± 2.2% → +1.7 ± 1.0% |
| Leg-press 1RM | +22 ± 14% (164 → 200 kg) | no change (158 → 159 kg) | interaction p < 0.001, ηp² = 0.62 |
| Plantar-flexor isometric force | +14 ± 14% (1177 → 1327 N) | no change | interaction p = 0.02, ηp² = 0.21 |
| Countermovement jump height | +5.9 ± 8.2% (0.33 → 0.35 m) | no change | interaction p = 0.002, ηp² = 0.32 |
| Countermovement jump peak power | +7.5 ± 7.3% (45.0 → 48.4 W/kg) | small fall (44.7 → 43.0, p = 0.04) | interaction p < 0.001, ηp² = 0.44 |
| Body composition | whole-body lean mass +1 ± 2% (p = 0.03), fat mass −11.5 ± 16.0% (p = 0.01), leg lean mass unchanged | no change | lean mass interaction p = 0.04; leg lean mass p = 0.32; body mass p = 0.74 |
Main results: oxygen cost at minute 90 changed −2.1% with strength vs +0.6% without (p = 0.04); in the strength group the drift from minute 15 to 90 fell from 4.7% to 2.1%; time to exhaustion afterwards rose 35% vs no change (interaction p = 0.004); fresh economy did not change in either group.
Two measures of "durability", not one. The headline −2.1% vs +0.6% is the change from before to after training in the oxygen cost AT minute 90, compared between groups. The drop in drift (4.7% → 2.1% from minute 15 to minute 90) is a within-group result in E+S only; the paper gives no between-group test of drift on that scale. Because minute-15 economy did not move, both point the same way.
Baseline drift was not equal. From the group means in Table 4 (our calculation), E+S drifted ~5.1% from minute 15 to 90 before training and E ~2.8%. Part of the E+S improvement may be regression from a high starting drift. After training the groups drifted ~2.3% vs ~4.3%.
The control group got slightly worse. In E, time to exhaustion fell (not significant), lactate at 90 min rose (2.4 → 2.9 mmol/L, p < 0.05) and RPE at 90 min rose non-significantly (16.5 → 17.3). A worsening control inflates every between-group contrast. The paper does not explain it; running volume fell a little in both groups.
Size in practical units. +77 s on a run to exhaustion of about 4-5 min at ~16 km/h. The authors note that earlier strength trials found +4-7% in time trials after long moderate exercise, and that time-to-exhaustion tests show much larger changes than time trials; +35% does not translate to a 35% faster finish.
Intensity of the 90 min. 13.1 km/h is 85% of the runners' 10 km speed. Our calculation: a Riegel prediction from a 39:02 10 km gives a ~3:00 marathon (~14.1 km/h), so the test run was ~7% slower than predicted marathon pace. Riegel tends to overpredict marathon speed in runners on modest mileage, so the true gap is probably smaller. Lactate sat at ~1.8-2.4 mmol/L through minute 60 and at 2.2-2.9 mmol/L at minute 90.
Mechanisms
Measured: less rise in oxygen cost (and energy cost) late in the run, lower late lactate and RPE in E+S relative to E, no difference in heart rate or RER. Strength and jump measures rose 6-22% while leg lean mass did not change.
Measured: the VO2 reached at exhaustion did not change in either group, and lactate at exhaustion rose in E+S. Authors' interpretation: the longer time to exhaustion was not a larger oxygen capacity in the fatigued state.
Authors' interpretation (not measured here; from cycling studies): a shift from type IIx to more efficient type IIa fibres matters most late in a run, when type I fibres tire as glycogen falls and type II fibres contribute more; strength training may also blunt the rise in neural activation (EMG) during long exercise; stiffer muscle-tendon units may help. The authors stress that these mechanisms "were not measured in the current investigation".
Authors' interpretation: the time-to-exhaustion gain may come from a larger D′ (work capacity above critical speed), from stronger muscles recruiting fewer motor units at the same speed, and from more glycogen left at minute 90 thanks to the smaller drift. None of these was measured; leg lean mass did not rise.
Authors' interpretation: the gains in strength without leg lean mass point to mainly neural adaptation.
Inference: part of the effect may be the higher starting drift in E+S (regression to the mean) and the slight decline in E; with 14 per group neither can be ruled out.
Practical Application
**Does lifting help economy hold up late in a long run?One small RCT (28 well-trained men, completers only) says it may: 10 weeks of heavy lifting plus jumps cut the oxygen-cost rise over a 90-min run from 4.7% to 2.1%, fresh economy unchanged, and a run to exhaustion straight after lasted 35% longer (247 → 324 s), while the running-only group did not improve.** Twice-weekly squats, single-leg press, calf work and jumps; the benefit showed up late, not in a fresh test.The run (~80% VO2max, just above the first lactate threshold) was at or a little below these runners' likely marathon pace (our calculation: ~7% slower than a Riegel prediction, which tends to flatter runners on ~45-50 km a week). The finish test was an open-ended run to exhaustion, which magnifies small changes, not a race.The 14 men who lifted had done no lower-limb strength work in the previous 6 months and trained 0.8 h a week more. It supports adding heavy lifting for a marathon or half; it does not promise a time.
Copy the programme, not just the idea: 2 supervised sessions a week, 2-4 days apart, ~45 min; one vertical and one horizontal jump drill (pogo jumps and hop-and-stick, then drop jumps and bounds); back squat and single-leg press 3 × 6-8 → 5-6 → 4-5 at 65-80 → 80-85 → 85-90% 1RM; heavy isometric calf raises.Squats were lifted as fast as possible, with the load set by bar speed.That matches the base's heavy default for runners (SYNTH-44); the jumps started sooner than the base's cautious ladder in SYNTH-45 — hop-and-stick from week 1, drop jumps and bounds from week 4 — under supervision with ground-contact feedback, with one training-related injury among 38 enrolled (group not stated).
Judge a strength block late in a long run, not fresh. A fresh economy test or a short time trial may show nothing after 10 weeks (as here) while the late-run cost improves. Inference: heart rate did not change in this trial (p = 0.83), so pace-to-heart-rate drift may not show the effect; the late-run signals that moved were oxygen cost, lactate and RPE.In the field, compare RPE over the last 30 min of comparable long runs at the same pace before and after the block, and read a flat heart-rate drift as no evidence either way.
Strength is added to running, not swapped for it. Running volume stayed roughly level (E+S 3.5 → 3.2 h a week), and regular long runs go with better economy durability (REF-160, cross-sectional, simple correlations). Inference: do not cut the long run to make room for the gym; this trial did not test replacing running with lifting.
Women and masters: untested here (men only, mean age ~35). The only earlier trial in women (female duathletes, Vikmoen 2017, not in the base) is described inconsistently in this paper — "no changes in performance or RE" in the Introduction, "similar results" in the Limitations; its economy over a 90-min run at 60% VO2max did not change.
Limitations
Small: 14 per group, completers only (28 of 38 enrolled), dropouts not split by group; no sample-size calculation, trial registration or blinding reported.
Endurance training was habitual and self-reported, not prescribed; volume did not differ between groups, but intensity was not reported and E+S trained 0.8 h a week more. There was no extra activity for the runners who only ran.
Group means were unequal before training in drift (~5.1% vs ~2.8%; our calculation) and time to exhaustion (247 vs 194 s), though the paper reports no significant baseline differences, and the control group got slightly worse; both inflate the contrast.
Outcomes: lab treadmill economy over 90 min and an open-ended run to exhaustion. No time trial, race or run longer than 90 min; the run was at or a little below likely marathon pace (our calculation).
Fresh economy did not improve, unlike most runner strength trials (the authors cite 3-8%); whether that reflects this sample or the programme is unknown.
Men only, Tier 2, no lower-limb strength training in the previous 6 months: the gains may be smaller in runners who already lift.
The headline economy result rests on a post hoc between-group difference at one time point (p = 0.04). Mechanisms were not measured.
Read from the accepted manuscript in the Loughborough repository; the copy-edited version may differ in detail. The abstract gives VO2max 58.6, while Table 2's group means are 58.9 and 56.8.
Funded by the National Strength and Conditioning Association; the conflict-of-interest statement lists only this grant.
How this entry was checked and changed
Why this level
parallel RCT, n=28 analysed (14 + 14) of 38 enrolled; read from the accepted manuscript (full text). It sits at the low edge of B+: completers only (10 dropouts, 26%, split by group not reported, one injury caused by the training), no blinding, trial registration or sample-size calculation reported, running self-reported and not standardised, and the strength group trained 0.8 h a week more with no matched extra activity in the runners who only ran
Full text (checked 2026-10-16)
Out of 38 participants that initially enrolled, 28 completed the study (E+S n=14; E n=14); dropouts: illness (n=2), voluntary withdrawal (n=1), no response (n=3), injury due to the training intervention (n=1), injury unconnected (n=4) (full text).
Maximal oxygen uptake 58.6 ml·kg-1·min-1; 10 km 39:02 mm:ss; 90 min run at 10% Δ between lactate threshold 1 and 2 (13.1±1.4 km/h, 79.7% V̇O2max); TTE at 95% pre-test V̇O2max (16.1±1.6 km/h) (full text).
Strength sessions separated by 2-4 days, ~45 min; 2 plyometric exercises and 3 lower limb strength exercises; bar speed between 0.35 and 0.55 m/s, 80-90% of 1RM; leg press 1RM re-tested at the beginning of week 6 (full text).
Table 1: back squat and single-leg press 3 x 6-8 / 120 s at 65-80 %, 3 x 5-6 / 120-150 s at 80-85 %, 3 x 4-5 / 150 s at 85-90 %; seated isometric calf raises 4 x 6-8 / 60 s at 80-100 %, then 5 x 4-6 / 90 s at 100 % (full text).
RE: training x group x run time F=3.81, p=0.003, ηp2=0.13; between-group difference at 90 min (-2.1 ± 3.5% (E+S) vs +0.6 ± 3.8% (E); p=0.04) (full text).
Relative to 15 min, within S+E: 75 min 3.4 ± 1.0 vs 1.5 ± 1.9% (p=0.003), 90 min 4.7 ± 1.7 vs 2.1 ± 1.0% (p<0.001); E no within-group interaction (p=0.42) (full text).
No training x group effects for RE at 10 km·h-1 (p=0.86), 12 km·h-1 (p=0.89) or after 15 min (F=1.03, p=0.32) (full text).
TTE: F=9.92, p=0.004, ηp2=0.28; E+S 35 ± 25% (247 ± 94 vs 324 ± 105 s, p<0.001); E 194 ± 96 vs 169 ± 106 s, p=0.40; abstract: +35% in E+S and -8% in E (full text).
BLa at end of TTE 5.6 ± 2.0 vs 6.4 ± 2.4 (p=0.04) in E+S, 4.9 ± 1.8 vs 5.3 ± 2.0 (p=0.39) in E; V̇O2 reached during TTE no effect (p=0.97) (full text).
Leg press 1 RM +22 ± 14% (ηp2=0.62), PF MVF +14 ± 14%, CMJ height +5.9 ± 8.2%, CMJ PP +7.5 ± 7.3%; whole body FM -11.5 ± 16.0% (p=0.01) (full text).
Running training volume F=0.10, p=0.76; total training volume F=3.70, p=0.07; E+S +22% (0.8 ± 1.1 h·week-1) vs E (-0.1 ± 1.5 h·week-1); adherence 99% (full text).
Funding: National Strength Conditioning Association "Young Investigator Research Grant" (full text).