Physical activity is associated with a reduced risk of colon, breast, endometrial, lung, and pancreatic cancer. Evidence for mediating molecular mechanisms from experimental studies substantially strengthens the causal inference for this relationship. Randomized controlled trials indicate that exercise affects metabolic profiles, including hormone levels (estrogen, insulin signaling), inflammation (e.g., C-reactive protein), and adipokine concentrations (e.g., leptin). The size of the effect depends frequently on concurrent changes in body composition. There is also initial evidence for effects on immune function, oxidative stress, and possibly DNA repair capacity. Finally, outdoor physical activity can directly increase 25(OH)-vitamin D levels, providing another potential mechanism for linking physical activity to cancer risk. Randomized controlled studies with biomarker measurements are essential to increase evidence for causality and to identify the most effective intervention strategies and pharmacologic targets.