Epidermal cell kinetics and DNA adduct levels, and skin morphological changes were measured following weekly topical applications for 29 weeks of high (16, 32 and 64 micrograms) benzo[a]pyrene (B[a]P) doses to female ICR/Harlan mice, in order to investigate the relationship of these parameters to the timing, incidence and morphology of the elicited tumors. During the tumor latency period, [3H]thymidine labeling index, mitotic index, epidermal cell stacking, incidence of pyknotic and dark basal keratinocytes and labeled mitoses were periodically measured, as were nuclear area and DNA content. DNA adducts in skin epidermis were measured by an ELISA method over a period of 9 weeks of single weekly applications of 64, 32, 16 or 8 micrograms B[a]P. There was an initial linear increase in DNA adducts with dose in the epidermis but the increase was much less steep above 32 micrograms/week. This did not correlate with the sharp rise in tumor response above the 32 micrograms/week dose rate. Cell kinetic changes in response to the 64 micrograms/week dose reached a plateau in the first few weeks of the tumor latent period. There was little epidermal hyperplasia but an associated dose-dependent increase in [3H]thymidine labeling index, mitotic index and incidence of pyknotic and dark cells. This evidence indicated that B[a]P produced extensive cytotoxicity and cell death with regenerative proliferation under these conditions. Giant keratinocytes occurred in all dose groups. Analysis of a labeled mitosis curve indicated that B[a]P produced a G2/M block. There was a marked inflammatory response in the dermis at all B[a]P doses. Mice were observed weekly for tumor formation. Virtually all of the tumors were papillomas on initial appearance and required an average of 8 weeks to convert to carcinomas. The substantial cell killing and regenerative proliferation, and the correspondence between the dose-response patterns for epidermal damage and tumors, together with the initial appearance of tumors in the benign form, a characteristic of the action of promoting agents, provided evidence that the tissue damage associated with the high dose levels of B[a]P used in this study reflected tumor-promoting activity in this mouse epidermal tumorigenesis model. The implication of the results for mathematical models of tumor formation are discussed.