Purpose: This study examined the relative influence of anatomical and neuromuscular variables on maximal isometric and concentric knee extensor torque and provided a comparative dataset for healthy young males. Methods: Quadriceps cross-sectional area (CSA) and fascicle length (l_f) and angle (θ_f) from the four quadriceps components; agonist (EMG:M) and antagonist muscle activity, and percent voluntary activation (%VA); patellar tendon moment arm distance (MA) and maximal voluntary isometric and concentric (60° s⁻¹) torques, were measured in 56 men. Linear regression models predicting maximum torque were ranked using Akaike’s Information Criterion (AIC_c), and Pearson’s correlation coefficients assessed relationships between variables. Results: The best-fit models explained up to 72 % of the variance in maximal voluntary knee extension torque. The combination of ‘CSA + θ_f + EMG:M + %VA’ best predicted maximum isometric torque (R² = 72 %, AIC_c weight = 0.38) and ‘CSA + θ_f + MA’ (R² = 65 %, AIC_c weight = 0.21) best predicted maximum concentric torque. Conclusion: Proximal quadriceps CSA was included in all models rather than the traditionally used mid-muscle CSA. Fascicle angle appeared consistently in all models despite its weak correlation with maximum torque in isolation, emphasising the importance of examining interactions among variables. While muscle activity was important for torque prediction in both contraction modes, MA only strongly influenced maximal concentric torque. These models identify the main sources of inter-individual differences strongly influencing maximal knee extension torque production in healthy men. The comparative dataset allows the identification of potential variables to target (i.e. weaknesses) in individuals.