Two short-term in vitro tests for mutagenicity (Salmonella reverse mutation and BHK21 cell transformation) were conducted on a series of fluorocarbons. Some of these materials (FC22, FC31, FC142b, FC143, and FC143a) were found to be positive in one or both of the tests and could therefore be considered as being potentially carcinogenic to animals. Such activity was not anticipated for what were previously considered inert materials and in consequence several examples of these fluorocarbons, which represented different combinations of short-term test results, were tested for carcinogenicity in limited in vivo bioassays. In these studies, rats were dosed for 1 year by gavage 5 days a week with either FC22, FC31, FC133a, FC134a, or FC143a dissolved in a corn-oil at a single dosage of 300 mg/kg body weight. The animals were then observed until week 125 with detailed necropsy at termination. The study revealed that FC31 was a potent carcinogen (to the rat stomach), a result which reflected the short-term test predictions, but FC133a, which gave a negative response in both the in vitro assays, induced a high incidence of reproductive tract tumors. The weak bacterial mutagens FC22 and FC143a did not induce tumors in this study, and the nonmutagenic FC134a was without overt carcinogenic activity. It is concluded that, while recognizing the limitations of the in vivo component of this study, the short-term tests were only partially successful in identifying potential carcinogens for this series of chemicals. Fluorocarbon 31 was a potent carcinogen which was first identified by bacterial mutation and cell transformation, whereas the equally potent carcinogen FC133a was not so identified. The lack of genotoxic activity with this particular compound leads us to believe that the carcinogenic activity may be due to mechanisms other than those which involve direct DNA interactions.