One hundred and sixty-nine patients with knee injuries were divided into isometric and isotonic exercise groups by a process of random selection. The groups were comparable with respect to age and type of injury suffered. The isometric group required a mean of 11.7 treatments and the isotonic group 14.9 treatments to return to duty. The difference between the two means, 3.2 treatments, is statistically and clinically significant. In both groups there was a small but statistically significant cross-education effect on the unexercised limb. Muscle hypertrophy in both groups was insignificant during the exercise period.