The force that healthy humans generated against a fixed pedal was measured and compared with that predicted by four models. The participants (n = 11) were seated on a stationary bicycle and performed brief pushing efforts against an instrumented pedal with the crank fixed. Pushes were performed to 10 force magnitude targets and at 12 crank angles. The increasing magnitude portion of the sagittal-plane force path for each push effort was fitted with a line to determine the direction of the muscle component of the foot force. Those directions varied systematically with the position of the pedal (crank angle) such that the force path lines intersected a common region superior and slightly anterior to the hip. The ability of four models to predict force path direction was tested. All four models captured the general variation of direction with pedal position. Two of the models provided the best performance. One was a musculoskeletal model consisting of nine muscles with parameters adjusted to provide the best possible ft. The other model was derived from (a) observations that the lines-of-action of the muscle component of foot force tended to intersect in a common region near the hip, and (b) the corresponding need for foot force to intersect the center-of-mass during walking. Thus, this model predicted force direction at each pedal position as that of a line intersecting the pedal pivot and a common point located near the hip (divergent point). The results suggest that the control strategy employed in this seated pushing task reflects the extensive experience of the leg in directing force appropriately to maintain upright posture and that relative muscle strengths have adapted to that pattern of typical activation.