The retrovirus XMRV (xenotropic murine leukemia virus–related virus) has been detected in human prostate tumors and in blood samples from patients with chronic fatigue syndrome, but these findings have not been replicated. We hypothesized that an understanding of when and how XMRV first arose might help explain the discrepant results. We studied human prostate cancer cell lines CWR22Rv1 and CWR-R1, which produce XMRV virtually identical to the viruses recently found in patient samples, as well as their progenitor human prostate tumor xenograft (CWR22) that had been passaged in mice. We detected XMRV infection in the two cell lines and in the later passage xenografts, but not in the early passages. In particular, we found that the host mice contained two proviruses, PreXMRV-1 and PreXMRV-2, which share 99.92% identity with XMRV over >3.2-kilobase stretches of their genomes. We conclude that XMRV was not present in the original CWR22 tumor but was generated by recombination of two proviruses during tumor passaging in mice. The probability that an identical recombinant was generated independently is negligible (~10–12); our results suggest that the association of XMRV with human disease is due to contamination of human samples with virus originating from this recombination event.