In order to achieve the general exercise objectives of increasing i) strength, ii) stamina, and iii) cardiovascular workout, a proper exercise regime has to achieve certain tradeoffs specific to the individual's muscle mechanics and the fatigue states. An adaptive approach is proposed to control a 1 degree of freedom (DOF) exercise machine in which the exercise regime is modified according to the estimates of the force-velocity relationship (Hill's curve) of the muscle. A optimal velocity profile is determined from the Hill's curve via a class of power related criteria. The adaptive controller simultaneously identifies the Hill's curve, and causes the user to operate at the estimated optimal velocity profile. The controller always interacts with the user passively. It is believed that the adaptive controller is able to modify the exercise regime accordingly as fatigue sets in. Stability and convergence results have been rigorously proven and in here some preliminary experimental results are presented.