Effect of Strength Training Programs in Middle- and Long-Distance Runners' Economy at Different Running Speeds: A Systematic Review with Meta-analysis
This systematic review with meta-analysis compared strength-training methods for running economy in middle- and long-distance runners, and asked whether the effect depends on the speed at which economy is measured.
Against controls who ran without strength training (or lifted below 40% of 1RM), high-load training (≥ 80% of 1RM) produced a small improvement in economy (ES = -0.266) and combined methods a moderate one (ES = -0.426); plyometric training improved economy only at speeds ≤ 12.00 km/h (ES = -0.307), and submaximal-load and isometric training produced no significant improvement (for isometric training, three small calf-only studies whose pooled estimate favoured training but whose confidence interval crossed zero; see Key Results).
The high-load effect was larger at higher running speeds and in runners with a higher VO2max.
Participants
Middle- and long-distance runners, no restriction on sex or level: 195 moderately trained, 272 well trained and 185 highly trained athletes across the included studies. 31 studies (full text).
Protocol
Search: Web of Science, PubMed, SPORTDiscus and SCOPUS, all articles indexed up to November 2022.
PICOS: strength training for ≥ 3 weeks, by method: high loads (≥ 80% of 1RM), submaximal loads (40-79% of 1RM), plyometric, isometric, combined (two or more methods); comparator = endurance running without strength training or with low loads (< 40% of 1RM); outcome = running economy before and after; randomised and non-randomised controlled studies.
Certainty of evidence by GRADE; three-level random-effects meta-analysis and moderator analysis in R (version 4.2.1).
Included programmes lasted between 6 and 24 weeks, with one to four sessions a week.
Key Results
| Method | Effect on running economy | Speeds assessed | GRADE certainty |
|---|---|---|---|
| High load (≥ 80% 1RM) | small, ES = -0.266, p = 0.039 | 8.64 to 17.85 km/h | moderate |
| Combined methods | moderate, ES = -0.426, p = 0.018 | 10.00 to 14.45 km/h | low |
| Plyometric | small, ES = -0.307, p = 0.028, only at ≤ 12.00 km/h (β1 = 0.470, p = 0.017) | — | moderate |
| Submaximal load (40-79% 1RM) | no improvement (p > 0.131) | 9.75 to 16.00 km/h in the Results (p = 0.131); 10.00 to 15.28 km/h in the abstract (corrected 2026-10-12, full text) | moderate |
| Isometric | no significant improvement (ES = -0.269, 95% CI -0.79 to 0.252, p = 0.253; 3 studies, 28 vs 27 participants, all isometric calf work) | 10.00 to 15.28 km/h in the Results (p = 0.253); 9.75 to 16.00 km/h in the abstract (corrected 2026-10-12, full text) | moderate |
Moderators of the high-load effect: running speed (β1 = -0.117, p = 0.027) and VO2max (β1 = -0.040, p = 0.020), i.e. greater improvements at higher speeds and in runners with a higher VO2max.
Negative effect sizes mean a lower energy cost, i.e. better economy.
Practical Application
Across controlled trials in 195 moderately trained, 272 well-trained and 185 highly trained middle- and long-distance runners, heavy strength training (≥ 80% of 1RM) improved running economy by a small amount (ES = -0.266) and combined methods by a moderate amount (ES = -0.426), plyometrics helped only at speeds ≤ 12.00 km/h (ES = -0.307), and submaximal-load and isometric training did not significantly help; the heavy-load effect grew with running speed and with VO2max. To "what kind of gym work makes me more economical?": lift heavy; a mix of methods did best, and light-load circuits or isometrics alone were not shown to be effective (the isometric pool was three small calf-only studies; two of them, REF-1155 and REF-1157, reported lower running cost after isometric training, though in REF-1157 the control group improved too).
For a fast runner with a high VO2max, heavy lifting is the method with the evidence for economy at race speed; plyometrics' benefit was confined to slower speeds.
Inference, not tested here: a slower recreational runner whose training and race speeds sit at or below 12 km/h may get as much from plyometrics as from heavy work, but pairing the two (a combined programme) had the largest pooled effect.
This meta-analysis measures economy, not race time; for the dose (sessions, weeks, in-season) see SYNTH-44.
Limitations
Non-randomised controlled studies were included; GRADE certainty was moderate at best and low for combined methods.
Economy was measured at laboratory test speeds, which differed by method, so the speed comparison across methods is indirect.
Full text read 2026-10-12 (see Full text); the isometric pool is three small calf-only studies (28 vs 27 participants).
How this entry was checked and changed
Why this level
meta-analysis of controlled trials that include non-randomised ones; GRADE certainty moderate for four methods and low for combined methods; 31 included studies
Full text (checked 2026-10-12)
31 studies were included (high load 11 studies/19 ESs, plyometric 11/28, submaximal 3/7, isometric 3/7, combined 9/20), with 652 participants (472 males and 180 females) aged between 17 and 45 years (full text).
High-load improvements came from interventions of 6 to 14 weeks at 2–4 days per week, and weeks, sessions per week and total sessions did not moderate the effect (all p > 0.111) (full text).
In high-load subgroups, the effect was −0.713 at > 12.00 km/h (p = 0.005) vs −0.060 at ≤ 12.00 km/h (p = 0.652), and −0.651 in highly trained (p = 0.006) vs 0.018 in moderately trained runners (p = 0.930) (full text, Table 5).
High-load effect by sex: male groups −0.360 (p = 0.041), female groups (2 groups, 2 ESs) 0.203 (p = 0.162) (full text, Table 5).
The main plyometric analysis was not significant (ES = −0.122, p = 0.167, speeds 7.00 to 18.00 km/h); the −0.307 effect is the ≤ 12.00 km/h subgroup, reported with p = 0.010 in the table (full text, Table 6).
Combined methods gave ES = −0.647 (p = 0.013) with significant heterogeneity before two outlier groups were removed, leaving −0.426 (p = 0.018); most combined studies included submaximal-load training (full text).
All three isometric studies used isometric ankle plantarflexion at ≥ 80% of maximal voluntary contraction, with muscle action times of 3 s in two studies and 20 s in the one that found no improvement (full text).
Correction: the entry lists the speeds assessed as 10.00 to 15.28 km/h for submaximal load and 9.75 to 16.00 km/h for isometric (as in the abstract), but the Results section reports submaximal load at 9.75 to 16.00 km/h (p = 0.131) and isometric at 10.00 to 15.28 km/h (p = 0.253) (full text).
Revision 2026-10-16: "isometric: no improvement" reworded as no significant improvement, with the pooled estimate from the Results (ES = -0.269, 95% CI -0.79 to 0.252, p = 0.253, Q(6) = 3.276) and the GRADE table (3 studies, 28 vs 27 participants).The three isometric studies are Albracht 2013 REF-1155 (90% MVC, 5 × 4 × 3 s, 14 weeks), Lum 2023 REF-1157 (maximal, 3 × 3 × 3 s, 6 weeks) and Fletcher 2010 (80% MVC, 3 × 4 × 20 s, 8 weeks, highly trained; not in the base); in the review's own reading only Fletcher showed no economy gain.Bohm 2021 REF-1156, which repeated Albracht's programme in recreational runners (energy cost -4%, within group), is neither among the 31 included nor among the 42 full-text exclusions; the review cites it only in its Discussion (reference 121), as one of two 14-week isometric studies that an earlier meta-analysis had added to its heavy-load analysis, and does not say why it was not pooled.Stale abstract-era lines (study count, Limitations) updated. Cross-references added.