A bioassay of titanium dioxide for possible carcinogenicity was conducted by administering the test chemical in feed to Fischer 344 rats and B6C3F1 mice. Groups of 50 rats of each sex and 50 mice of each sex were administered titanium dioxide in the diet at one of two doses, either 25,000 or 50,000 ppm, for 103 weeks and then observed for 1 additional week. Matched controls consisted of 50 untreated rats of each sex and 50 untreated mice of each sex. All surviving rats and mice were killed at 104 weeks. Administration of the titanium dioxide had no appreciable effect on the mean body weights of rats or mice of either sex. With the exception of white feces, there was no other clinical sign that was judged to be related to the administration of titanium dioxide. Survival of the rats and the male mice at the end of the bioassay was not affected by the test chemical; mortality in female mice was dose related. Sufficient numbers of dosed and control rats and mice of each sex were at risk for development of late-appearing tumors. In the female rats, C-cell adenomas or carcinomas of the thyroids occurred at incidences that were dose related (P=0.013), but were not high enough (P=0.043 for direct comparison of the high-dose group with the control group) to meet the level of P=0.025 required by the Bonferroni criterion (controls 1/48, low-dose 0/47, high-dose 6/44). Thus, these tumors of the thyroid were not considered to be related to the administration of the test chemical. In male and female mice, no tumors occurred in dosed groups at incidences that were significantly higher than those for corresponding control groups. It is concluded that under the conditions of this bioassay, titanium dioxide was not carcinogenic by the oral route for Fischer 344 rats or B6C3F1 mice.