This study aimed to determine the specific adaptations provoked by poweroriented resistance training using light (LL-PT, 40% 1-RM) vs. heavy (HL-PT, 80% 1-RM) loads in older adults. Using a randomized within-subject study design, 45 older adults (>65 years) completed an 8-week control period (CTR) followed by 12 weeks of unilateral LL-PT vs. HL-PT on a leg press. The 1-RM, theoretical force at zero velocity (F 0 ), maximal unloaded velocity (V 0 ), and maximal muscle power (P max ) were determined through a force-velocity relationship test. Isometrically, the rate of force development (RFD) and the corresponding muscle excitation of the knee extensor muscles were assessed. In addition, muscle cross-sectional area (CSA) and architecture of two quadriceps muscles were determined. Changes after CTR, LL-PT and HL-PT were compared using linear mixed models. HL-PT provoked greater improvements in 1-RM and F 0 (effect size (ES) = 0.55-0.68; p < 0.001) than those observed after LL-PT (ES = 0.27−0.47; p ≤ 0.001) (post hoc treatment effect, p ≤ 0.057). By contrast, ES of changes in V 0 was greater in LL-PT compared to HL-PT (ES = 0.71, p < 0.001 vs. ES = 0.39, p < 0.001), but this difference was not statistically significant. Both power training interventions elicited a moderate increase in P max (ES = 0.65-0.69, p < 0.001). Only LL-PT improved early RFD (ie, ≤100 ms) and muscle excitation (ES = 0.36-0.60, p < 0.05). Increased CSA were noted after both power training programs (ES = 0.13-0.35, p < 0.035), whereas pennation angle increased only after HL-PT (ES = 0.37, p = 0.004). In conclusion, HL-PT seems to be more effective in improving the capability to generate large forces, whereas LL-PT appears to trigger greater gains in movement velocity in older adults. However, both interventions promoted similar increases in muscle power as well as muscle hypertrophy.