We have proposed that an early step in tumor progression is the expression of a mutator phenotype resulting from mutations in genes that normally function in the maintenance of genetic stability. There is new and strong experimental evidence that supports the concept of a mutator phenotype in cancer. As technologies for chromosomal visualization and DNA advance, there are increasing data that human cancer cells contain large numbers of mutations. First, I will review the concept of a mutator phenotype. Second, I will present the recent evidence that individual cancer cells contain thousands of mutations. Third, I will explore potential target genes that are required for maintenance of genetic stability in normal cells and ask if they are mutated in cancer cells. Fourth, I will address the timing of a mutator phenotype; is it an early event during tumor progression? Do tumors already contain cells that harbor mutations rendering them resistant to most chemotherapeutic agents? Lastly, I will speculate on the theoretical and practical implication of a mutator phenotype in cancer and consider the possibility of cancer prevention by delay, i.e., a reduction in mutation rates early during carcinogenesis might slow the progression of tumors.