It is estimated that up to 43 million Americans have a disabling condition that limits one or more of their daily activities, and that 38% of people with disabilities have significant mobility limitations. Biomechanically-based movement analysis is used increasingly in the surgical and medical treatment of a variety of neuromuscular and musculoskeletal disorders associated with these functional limitations. Inverse dynamics analysis provides data on the joint kinetics -- net muscle moments and powers used at joints during movement -- which reveal the neuromuscular strategies used to produce the observed movement. A patient's gait patterns are often evaluated using comparisons with data obtained from healthy individuals, despite the fact that many patients walk at much slower speeds. Kinetic patterns were investigated for 18 healthy adult subjects (9 male, 9 female) who walked at five speeds ranging from 25% to 125% of a normal walking speed. At the knee, the shapes of the mean patterns were similar for the three fastest, most natural walking speeds. At the two slowest speeds, however, subjects used little knee flexion and small knee moments through mid-stance, resulting in negligible joint power through the first 80% of the stance phase. At these speeds, most subjects had moment patterns that could not be classified with confidence as """"""""normal"""""""" or as predominantly flexor or extensor. Only four subjects had knee flexor moment patterns at the slowest speeds, and these included much smaller flexor moments in early stance phase than in some previous reports.