The current study was undertaken to investigate the dose-response characteristics of dieldrin-mediated enhancement of liver tumour formation in CF-1 mice. The median time to tumour development was established in controls, and in dieldrin-treated animals at six levels of continuous exposure (0.1, 1, 2.5, 5, 10 and 20 p.p.m.). The results of the analysis, which was based on liver tumour data from two parallel chronic feeding studies involving 1800 mice, are at variance with those reported by Druckrey for various established chemical carcinogens. In a double-logarithmic system of coordinates there was no linear relationship between the median total dose or the median time to tumour formation and the daily dieldrin exposure level. These results suggest that the tumourigenicity of this compound in CF-1 mouse liver is determined not by the sum of all consecutive doses, but rather by the level of daily exposure, and, presumably, the duration of treatment. This concept is consistent with the observed dose-dependency and reversible nature of dieldrin-induced subcellular changes in mouse liver. These considerations, together with evidence that dieldrin and its mammalian metabolites possess neither genotoxic activity nor potential, are not inconsistent with the concept that this compound is devoid of initiating potential, and operates by enhancing the effect of a genetically linked oncogenic factor in CF-1 mouse liver.