To determine its carcinogenic potential, sodium fluoride (NaF) was fed to CD-1 mice for up to 97 weeks. Mice given NaF at a dose of 4, 10, or 25 mg/kg of body weight per day added to a low-fluoride diet were compared to controls given either an unsupplemented low-fluoride diet or laboratory chow. Nonneoplastic changes consistent with those previously recognized from fluoride toxicity were observed in teeth, bones, and joints. Unexpectedly, osteomas occurred in all groups. The incidence of osteomas was similar in groups given the low-fluoride control diet, laboratory chow, or NaF doses of 4 or 10 mg/kg per day. The incidence of osteomas in these groups was increased over that historically experienced at the laboratory and reported in the literature for CD-1 mice. The incidence of osteomas in the mice given 25 mg NaF/kg per day added to a low-fluoride diet was increased over that in the other groups. Osteomas were first observed at Week 55. No malignant bone tumors were observed during the course of the study. The locations, multiplicity, and morphologic features of the osteomas in all groups were similar to those associated with virus-induced bone tumors. Electron microscopic examination revealed abundant retrovirus particles in all osteomas examined from control and test mice. It was concluded that the study was confounded by a retrovirus which contributed to the induction of the osteomas. Because the study was confounded, it cannot be considered a valid bioassay to be used for risk assessment.