Investigations using quadrupeds have suggested that the motor programs used for slope walking differ from that used for level walking. This idea has not yet been explored in humans. The aim of this study was to use electromyographic (EMG) signals obtained during level and slope walking to complement previously published joint angle and joint moment data in elucidating such control strategies. Nine healthy volunteers walked on an instrumented ramp at each of five grades (-39%, -15%, 0%, +15%, +39%). EMG activity was recorded unilaterally from eight lower limb muscles (gluteus maximus (GM), rectus femoris (RF), vastus medialis (VM), biceps femoris (BF), semimembranosus (SM), soleus (Sol), medial gastrocnemius (MG), and tibialis anterior (TA)). The burst onset, duration, and mean activity were calculated for each burst in every trial. The burst characteristics were then averaged within each grade and subject and submitted to repeated measures ANOVAs to assess the effect of grade (alpha=0.05, a priori). Power production increased during upslope walking, as did the mean activity and burst durations of most muscles. In this case, the changes in muscle activity patterns were not predictable based on the changes in joint moments because of the activation of biarticular muscles as antagonists. During downslope walking power absorption increased, as did knee extensor activity (mean and duration) and the duration of the ankle plantarflexor activity. The changes in muscle activity during this task were directly related to the changes in joint moments. Collectively these data suggest that the nervous system uses different control strategies to successfully locomote on slopes, and that joint power requirements are an important factor in determining these control strategies.