Typical petroleum hydrocarbon mixtures were tested directly, without extraction, in the Salmonella/microsome mutagenesis assay in order to determine if the assay would be useful to predict their carcinogenic activity. The carcinogenic activity of each sample had been previously characterized in the in vivo mouse dermal carcinogenesis bioassay. The series of samples evaluated offered several advantages. They spanned a wide boiling point range, were well characterized chemically, had been tested for carcinogenic activity in a single laboratory, and varied in potency in vivo from inactive to highly active. Mutagenicity testing was performed in several well-established contract laboratories that routinely perform the assay. These laboratories were the main contracting laboratories for these assays at the time and had previously tested petroleum samples for clients. Initially, the first laboratory tested 13 samples in five strains of Salmonella typhimurium with and without rat liver S-9 (Arochlor 1254 induced), utilizing both plate and suspension techniques. None of the 13 samples exhibited a mutagenic response, even though 9 of the 13 were slightly to highly dermally carcinogenic in mice. Because of the unexpected results, it was decided to repeat the mutagenicity assays in two other laboratories. Six of the thirteen samples were selected, ranging in carcinogenic potency from negative to highly active. Again, none were mutagenic in the second contract laboratory. In a third facility, only one sample of the six exhibited a definite mutagenic response. However, the response was observed with a sample having only weak carcinogenic activity and, unusual for petroleum hydrocarbons, occurred without activation.(ABSTRACT TRUNCATED AT 250 WORDS)