To gain further insight into the neural mechanisms for different forms of quadrupedal walking, data on postural orientation, hindlimb kinematics, and motor patterns were assessed for four grades of downslope walking, from 25% (14 degrees slope) to 100% (45 degrees), and compared with data from level and downslope walking at five grades (5-25%) on the treadmill (0.6 m/s). Kinematic data were obtained by digitizing ciné film, and electromyograms (EMGs) synchronized with kinematic records were taken from 13 different hindlimb muscles. At grades from 25 to 75%, cycle periods were similar, but at the steepest grade the cycle was shorter because of a reduced stance phase. Paw-contact sequences at all grades were consistent with lateral-sequence walking, but pace walking often occurred at the steepest grades. The cats crouched at the steeper grades, and crouching was associated with changes in fore- and hindlimb orientation that were consistent with increasing braking forces and decreasing propulsive forces during stance. The average ranges of motion at the hindlimb joints, except at the hip, were often different at the two steepest slopes. During swing, the range of knee- and ankle-joint flexion decreased, and the range and duration of extension increased at the ankle joint to lower the paw downward for contact. During stance the range of flexion during yield increased at the ankle joint, and the range of extension decreased at the knee and metatarsophalangeal joints. Downslope walking was also associated with EMG changes for several muscles. The hip extensors were not active during stance; instead, hip flexors were active, presumably to slow the rate of hip extension. Although ankle extensors were active during stance, their burst durations were truncated and centered around paw contact. Ankle flexors were active after midstance at the steeper slopes before the need to initiate swing, whereas flexor and extensor digit muscles were coactive throughout stance. Overall the changes in posture, hindlimb kinematics, and activity patterns of hindlimb muscles during stance reflected a need to counteract external forces that would accelerate angular displacements at some joints. Implications of these changes are discussed by using current models for the neural control of walking.