The purpose of this study was to determine the effects of two different strength-power training models on sprint performance. Forty-eight soldiers of the Brazilian brigade of special operations with at least one year of army training experience were divided into a control group (CG: n = 15, age: 20.2 ± 0.7 years, body height: 1.74 ± 0.06 m, and body mass: 66.7 ± 9.8 kg), a traditional training group (TT: n = 18, age: 20.1 ± 0.7 years, body height: 1.71 ± 0.05 m, and body mass: 64.2 ± 4.7 kg), and a complex training group (CT: n = 15, age: 20.3 ± 0.8 years, body height: 1.71 ± 0.07 m; and body mass: 64.0 ± 8.8 kg). Maximum strength (25% and 26%), CMJ height (36% and 39%), mean power (30% and 35%) and mean propulsive power (22% and 28%) in the loaded jump squat exercise, and 20-m sprint speed (16% and 14%) increased significantly (p<0.05) following the TT and CT, respectively. However, the transfer effect coefficients (TEC) of strength and power performances to 20-m sprint performance following the TT were greater than the CT throughout the 9-week training period. Our data suggest that TT is more effective than CT to improve sprint performance in moderately trained subjects.