REF-1719 · Li et al., 2026 · Scandinavian Journal of Medicine & Science in Sports, · Paper · PubMed

Resisted sprint training (sleds and similar loads) gave small gains in athletes' acceleration (g 0.32, 95% CI 0.16 to 0.49), vertical jump (0.25), horizontal jump (0.24) and change of direction (0.46), and small extra gains over unresisted sprint training (g 0.22-0.30, no intervals in the abstract), in a three-level meta-analysis of 31 controlled studies (179 effect sizes) from a review of 66 studies and 1945 athletes; GRADE certainty was moderate for acceleration and low for the rest, prediction intervals crossed zero for three outcomes, and no load, dose or modality stood out

Main caveat: The authors' own: benefit in new training contexts is uncertain (prediction intervals); no universal load, dose or superior modality.
fairly strong evidenceMeasured on trained to highly trained athletesRead from the abstract only

Li and colleagues asked whether resisted sprint training, used mainly to improve acceleration, also carries over to jumping and changing direction, and whether the effect depends on the comparator or on how the training is done.

They searched eight databases to July 2026 and included 66 independent studies with 1945 athletes; 31 controlled studies supplied 179 effect sizes to the main analysis, which used three-level random-effects models to handle several effect sizes per study.

Resisted sprinting favoured all four outcomes: acceleration (g 0.32), vertical jump (0.25), horizontal jump (0.24) and change of direction (0.46). Prediction intervals, which show the range of effects a new setting might see, crossed zero for acceleration, vertical jump and change of direction but not for horizontal jump.

Once comparator type was accounted for, the four domains did not clearly differ (p = 0.699). Over unresisted sprinting, resisted sprints added only small average effects (g 0.22-0.30). In exploratory moderator tests, neither comparator type nor external load, training dose or modality showed a clear link with the effect.

GRADE certainty was moderate for acceleration and low for the other outcomes. The authors conclude that resisted sprinting has favourable average effects in all four domains, that its benefit in a new setting is uncertain, that no load, dose or modality is shown to be best, and that it is best used as a targeted overload alongside fast sprinting and task-specific training.

Participants

66 independent studies, 1945 athletes; 31 controlled studies in the primary analysis.
Mostly male, by the authors' statement; sports, levels, ages and the female share are not in the abstract.

Protocol

Search: eight databases (not named in the abstract) to 7 July 2026.
Intervention: resisted sprint training; modalities, loads and doses varied (details not in the abstract).
Outcomes: acceleration, vertical jump, horizontal jump, change-of-direction performance.
Analysis: three-level random-effects meta-analysis (Hedges' g) accounting for dependence within studies; 95% confidence and prediction intervals; adjustment for comparator type; exploratory moderators (comparator type, external load, training dose, RST modality); GRADE certainty.
Not in the abstract: the risk-of-bias tool and its results, heterogeneity statistics, the comparators other than unresisted sprinting, and what the 35 studies outside the primary analysis contributed.

Key Results

Outcome (RST vs comparators)g (95% CI)Prediction interval crosses zeroGRADE
Acceleration0.32 (0.16 to 0.49)yesmoderate
Vertical jump0.25 (0.06 to 0.44)yeslow
Horizontal jump0.24 (0.11 to 0.37)nolow
Change of direction0.46 (0.09 to 0.82)yeslow
Domains compared after adjusting for comparator type: no clear difference (p = 0.699).
Against unresisted sprint training: additional mean effects small, g 0.22-0.30 (the abstract does not say which outcomes the range spans); intervals not given in the abstract.
Moderators (exploratory): no clear association with comparator type, external load, training dose or RST modality.

Practical Application

Resisted sprints are a reasonable add-on for acceleration, with small average gains (g 0.32) that may not appear in every squad. The prediction interval for acceleration crosses zero, so in a new setting the effect could be nil; the average result is at moderate certainty, the jump and change-of-direction results at low certainty.
No sled load, dose or type is shown to be best. The exploratory moderator analyses found no link with external load, dose or modality, so a specific percentage of body mass should not be quoted as a finding; REF-765's load ranges are what was tested, not a recommendation.
Where it differs from the guide: SYNTH-59 says, from REF-765, that sled sprinting has not been shown to beat the same sprints without a sled. This analysis found small additional mean effects over unresisted sprint training (g 0.22-0.30), with no intervals or certainty grade for that comparison in the abstract.That tilts the evidence slightly toward resisted sprints but does not settle it, and plain fast sprinting stays the core (REF-804; the authors also place resisted sprinting beside high-speed sprinting, not instead of it).
Transfer to jumping and change of direction is about as large as to acceleration (domains not clearly different after adjustment), so the method is not acceleration-specific in these data.
Inference, not tested here: for an endurance athlete who sprints only to sharpen (SYNTH-46), resisted sprints add little that hill sprints or plain sprints do not; the evidence is from athletes training for speed, mostly men.

Limitations

The authors' own: benefit in new training contexts is uncertain (prediction intervals); no universal load, dose or superior modality; a mostly male evidence base limits transfer to female athletes.
GRADE certainty is low for three of the four outcomes.
The risk-of-bias assessment, heterogeneity values and the intervals for the comparison with unresisted sprinting are not in the abstract.
Moderator analyses are exploratory and between studies.
Only 31 of 66 studies entered the primary analysis; the role of the rest, and the sports, levels and ages, are not described in the abstract. Full text not read; funding not known.
How this entry was checked and changed
Why this level

the abstract reports the GRADE certainty and what it found (moderate for acceleration, low for vertical jump, horizontal jump and change of direction), so the abstract-only cap does not apply; held at A- by the rubric because certainty is low for three of four outcomes, and because the prediction intervals cross zero for those three outcomes too; risk of bias for each study, heterogeneity values and the comparator-specific intervals are not in the abstract