The influence of parameter values (i.e., fiber optimum lengths and moment arms) and simplification of the geometry of a Hill-type muscle model on the prediction of normalized maximal isometric knee extension moment to knee joint angle relationship was studied. For that purpose, the geometry of m. quadriceps femoris was modeled in considerable detail, and all parameter values were determined on one set of cadaver specimens that had been selected for muscular appearance. The predicted relationship was compared to that measured in human subjects over the full range of physiological knee angles, and a good correspondence was found (r = .96). The good correspondence could be attributed to the substitution of realistic parameter values into the model. Incorporating complex muscle geometry into the model resulted in a small additional improvement of the prediction. It was speculated that the variation in results of cadaver measurements among studies reflects true differences caused by individuals' levels of physical activity in the period preceding death.Key Words: m. quadriceps femoris, muscle model, geometry, human cadavers Insight into the factors determining performance of motor tasks will benefit practitioners in athletics and rehabilitation. It is generally recognized that performance depends on properties of the musculoskeletal system and on control, with the effects of control being determined by the stimulation patterns that the nervous system sends to the muscles. In order to gain insight into the importance of system properties and control, it is necessary to systematically manipulate these factors and monitor the effects on performance. Unfortunately, system properties cannot be changed in living subjects for obvious reasons, and subjects are unable to systematically manipulate their control. For that reason, researchers have turned to models of the neuromusculoskeletal system. These models may be constructed such that the input consists of muscle stimulation patterns and the output is the movement and therewith the performance on the task (Levine,