Male rats were fed, from weaning onwards, either 2, 12.5 or 25% sunflower seed oil (polyunsaturated fatty acids, PSA) or lard (saturated fatty acids, SFA) and from the age of 15 weeks subgroups were given N-nitrosodimethylamine (NDMA) for 30 weeks. Blood levels of lipids were assayed and during the study exhaled ethane was measured as an index of in vivo lipid peroxidation (LPO). At the age of 50 weeks, rats were killed and livers were analysed for tumours. PSA diets decreased plasma cholesterol and triglyceride concentrations vs. respective SFA diet; NDMA administration did not affect plasma cholesterol but enhanced triglyceride concentration. NDMA markedly enhanced LPO. An increase in dietary fat content from 2 to 25% enhanced ethane exhalation, more in rats fed PSA than the SFA diet. In the 25% PSA group, indomethacin in the diet strongly inhibited LPO. Prevalence of liver haemangiosarcomas increased from 42% to 80% (p less than 0.05) in NDMA-treated animals when PSA increased from 2 to 25%; in the group having a 25% PSA diet containing indomethacin, the NDMA-induced tumour incidence was reduced to 64%. In NDMA-treated rats fed SFA diets the prevalence of haemangiosarcoma increased from 43% (2% fat) to 67% (25% fat). The data show that NDMA modifies plasma lipids and increases LPO. The quantity and saturation degree of fats altered the frequency of chemically-induced tumours and modified LPO. As an index of free radical reactions, LPO may have an important role in carcinogenesis. Dietary fat thus appears to promote carcinogenesis through mechanisms that involve LPO.