I have argued that most germline mutations are due to endogenous process. I speculate that endogenous control of germline mutation serves an important biological function. A certain rate of mutation is required to generate sufficient variation for adaptation during evolutionary time. Sexual reproduction and recombination may serve to enhance variation, but ultimately new germline mutation is required to replenish variant alleles lost secondary to negative selection, genetic drift, and population bottlenecks. Unfortunately, the requisite mutation rates carry a terrible price: for each advantageous mutation, there are many disadvantageous ones. Thus, all mammals are plagued with substantial Mendelian and multifactorial disease. Consequently, the optimal mutation rate should be at a level just sufficient to maintain the requisite variation needed for adaptation. In this view, mechanisms for negative selection are necessary to keep the mutation rate in check. If a high germline mutation rate produces a high mutation rate in somatic tissues, cancer may be an important mediator of negative selection. The multiple mutations necessary to produce cancer serve to amplify relatively small differences in the mutation rate, thereby providing an efficient selection against individuals with germline mutations that result in a high mutation rate. This hypothesis can account for the general similarity of the longevity-corrected cancer incidence profile and the small but significant prevalence of cancer before and during the reproductive period. While this hypothesis must presently be viewed as speculative, it integrates certain previously disjointed observations and suggests an alternative to the general assumption that cancer represents a breakdown in normal physiology.