Farm raised rainbow trout (Oncorhynchus mykiss) were exposed by various routes to benzo(a)pyrene (BP) as a representative carcinogenic polycyclic aromatic hydrocarbon (PAH). Following exposure of fish to the chemical by intraperitoneal (i.p.) injection, P-32-postlabelling studies indicated that non-feral trout were relatively resistant to the formation of BP-DNA adducts in liver. No adducts were detected in fish exposed to single doses (20 mg/kg) of BP. Multiple exposures (e.g. 2 x 25 mg/kg) were necessary in order for adducts to be detected, indicating that induction of the metabolising enzymes required for the bioactivation of BP is necessary. These studies provided reference information on DNA adducts for comparison with data from subsequent experiments at environmentally realistic low level exposures. Two types of low level aquatic exposure were carried out. The first procedure exposed fish for 30 days to a nominally constant low level (1.2 and 0.4 mu g/l) of a homogeneous dispersion of BP in water, to simulate low level aquatic environmental exposures. Following P-32-postlabelling analysis of the liver DNA of exposed fish, BP-DNA adducts were not detected. In the second procedure, fish were exposed to a constant low level of BP (ca. 0.5 mu/l) for 15 days then to a pulse (60 mu g/l) which was allowed to naturally decline (to ca. 2 mu g/l) during a further 15 days. Following this exposure, significant levels of BP-DNA adducts were detected in livers of trout. The effect of dietary exposures was investigated by feeding trout a diet containing either 58 mu g or 288 mu g BP per day for 6 days, equivalent to total doses of 43 mg/kg and 216 mg/kg. In both cases BP-DNA adducts were detected in livers of exposed fish. The results provide useful information on the types of exposures to PAHs which may pose a genotoxic risk to fish in the environment.