There have been no systematic comparisons of skeletal muscle adaptations in response to voluntary wheel running under controlled loading conditions. To accomplish this, a voluntary running wheel for rats and mice was developed in which a known load can be controlled and monitored electronically. Five-week-old male Sprague-Dawley rats (10 rats/group) were assigned randomly to either a 1) sedentary control group (Control); 2) voluntary exercised with no load (Run-No-Load) group; or 3) voluntary exercised with additional load (Run-Load) group for 8 wk. The load for the Run-Load group was progressively increased to reach approximately 60% of body weight during the last week of training. The proportions of fast glycolytic (FG), fast oxidative glycolytic (FOG), or slow oxidative (SO) fibers in the plantaris were similar in all groups. The absolute and relative plantaris weights were greater in the Run-Load group compared with the Control and Run-No-Load groups. The mean fiber cross-sectional areas of FG, FOG, and SO fibers were 20, 25, and 15% greater in the Run-Load than in Control rats. In addition, these fiber types were 16, 21, and 12% larger in Run-Load than in Run-No-Load rats. The muscle weights and mean cross-sectional areas of each fiber type were highly correlated with the average running distances and total work performed in the Run-Load, but not the Run-No-Load, group. The slope of the relationship between fiber size and running distance and total work performed was significant for each fiber type but was higher for FG and FOG fibers compared with SO fibers. These data show that the load on a rat running voluntarily can determine the magnitude of a hypertrophic response and the population of motor units that are recruited to perform at a given loading condition.