Abstract: Bachero-Mena, B and González-Badillo, JJ. Effects of resisted sprint training on acceleration with three different loads accounting for 5, 12.5, and 20% of body mass. J Strength Cond Res 28(10): 2954–2960, 2014—The optimal resisted load for sprint training has not been established yet, although it has been suggested that a resistance reducing the athlete's velocity by more than 10% from unloaded sprinting would entail substantial changes in the athlete's sprinting mechanics. This investigation has evaluated the effects of a 7-week, 14-session, sled-resisted sprint training on acceleration with 3 different loads according to a % of body mass (BM): low load (LL: 5% BM, n = 7), medium load (ML: 12.5% BM, n = 6), and high load (HL: 20% BM, n = 6), in young male students. Besides, the effects on untrained exercises: countermovement jump (CMJ), loaded vertical jump squat (JS), and full squat (SQ) were analyzed. The 3 groups followed the same training program consisting in maximal effort sprint accelerations with the respective loads assigned. Significant differences between groups only occurred between LL and ML in CMJ (p ≤ 0.05), favoring ML. Paired t-tests demonstrated statistical improvements in 0–40 m sprint times for the 3 groups (p ≤ 0.05), and in 0–20 m (p ≤ 0.05) and 0–30 m (p < 0.01) sprint times for HL. Sprint times in 10–40 m (p < 0.01) and 20–40 m (p ≤ 0.05) were improved in LL. Time intervals in 20–30 m and 20–40 m (p ≤ 0.05) were statistically reduced in ML. As regards, the untrained exercises, CMJ and SQ for ML and HL (p ≤ 0.05) and JS for HL were improved. The results show that depending on the magnitude of load used, the related effects will be attained in different phases of the 40 m. It would seem that to improve the initial phase of acceleration up to 30 m, loads around 20% of BM should be used, whereas to improve high-speed acceleration phases, loads around 5–12.5% of BM should be preferred. Moreover, sprint-resisted training with ML and HL would enhance vertical jump and leg strength in moderately trained subjects.