Purpose: The current study investigated the role of quadriceps and gastrocnemii size and "vastus lateralis" and "gastrocnemius medialis" muscle architecture in peak-power and time-to-peak-power exerted in an all-out Wingate test. Twenty-one amateur cyclists were recruited. Methods: Quadriceps and gastrocnemii anatomical cross-sectional area (ACSA), and "vastus lateralis" and "gastrocnemius medialis" pennation angle and fascicle length were measured using ultrasound. Relative peak-power (normalized per body mass) and time-to-peak-power were measured during a 30s all-out test. Results: Relative peak-power was correlated with quadriceps ACSA (r = 0.896, p < 0.001), gastrocnemii ACSA (r = 0.811, p < 0.001), "vastus lateralis" (r = 0.787, p < 0.001) and "gastrocnemius medialis" pennation angle (r = 0.638, p < 0.003). Multiple regression revealed that quadriceps and gastrocnemii ACSA accounted for 85% (R[superscript 2]= 0.85) of peak-power variance. Time-to-peak-power showed "very large" (r = -0.868, p < 0.001) and "large" correlation (r = -0.680, p = 0.001) with VL and GM fascicle length, respectively. Multiple regression analysis revealed that VL fascicle length explained 75% (R[superscript 2]= 0.75) of the time-to-peak-power variance. Conclusions: Quadriceps and gastrocnemii ACSA largely explained relative peak-power in an all-out Wingate test. "Vastus lateralis" fascicle length was the main predictor of the time-to-peak-power. Muscle architecture characteristics seem to be involved in the power generating capacity.