The effect of certain procarcinogens, among which demethylnitrosamine (DMN) is included, has been difficult to detect in several short-term assays. An alternative system, in which DMN effects could be easily quantitated, might be useful in studies of chemical carcinogenesis and environmental contamination. To develop such a system, we tested the possibility of measuring the amount of breakage produced by DMN on radiolabeled DNA of primary liver cultures. Rat liver cells were isolated 20 to 24 hr after partial hepatectomy, cultured, and pulse labeled in vitro with [3H]thymidine. Radioactively labeled cultures were treated with DMN or with the direct carcinogen N-methyl-N'-nitro-N-nitrosoguanidine and then lysed directly onto alkaline sucrose gradients. DMN and N-methyl-N'-nitro-N-nitrosoguanidine caused a dose-dependent reduction in the molecular weight of DNA, N-methyl-N'-nitro-N-nitrosoguanidine being approximately 1000 times more potent than DMN. DNA breaks appeared to be carcinogen specific and not due to cell death since treatment with high doses of cycloheximide, a noncarcinogenic hepatotoxic, was without significant effect. Our data indicate that detection of DNA breaks constitutes a more sensitive assay of DMN effects than does unscheduled DNA synthesis in primary liver cultures. Therefore, it could be useful to extend our work to determine the general applicability of quantitation of DNA breaks in liver cells as a short-term assay for the identification of possible carcinogens and procarcinogens.