Tumor progression to metastasis usually is assumed to occur through clonal genomic and epigenetic evolution. However, present evidence that challenges this paradigm. They show that genomic aberrations in tumor cells disseminated in the bone marrows of patients with no clinical evidence of metastasis generally do not resemble the aberrations in the primary tumors from which they arose. They interpret this to mean that tumor cells disseminate very early and evolve to metastatic disease independent from the primary tumor. Their model suggests that adjuvant therapies should be targeted to lesions in the disseminated cells rather than lesions found in primary tumors.