The induction of micronucleated erythrocytes by diisopropylcarbodiimide (DIC) and dicyclohexylcarbodiimide (DCC) was investigated as part of a U.S. National Toxicology Program (NTP) evaluation of the subchronic toxicity of these chemicals. Analysis of peripheral blood smears from male and female B6C3F1 mice exposed to 17.5-140.0 mg DIC/kg/day by skin painting for 13 weeks revealed dose-related increases in the frequency of micronucleated normochromatic erythrocytes (MN-NCE) in both sexes. Results of a similar 13-week peripheral blood micronucleus (MN) test with DCC (1.5-12.0 mg/kg/day) were also positive, although the increases in MN-NCE were not as great as those observed with DIC. In contrast to the positive results of the subchronic skin-painting studies in mice, acute bone marrow MN studies with DIC and DCC in male F344 rats, using intraperitoneal (i.p.) injection, yielded negative results. Both the acute and the subchronic exposures included doses that produced clinical signs of toxicity. Acute mouse bone marrow MN tests with DIC administered in single or triple i.p. injection protocols were subsequently conducted to determine if the differing responses between mice and rats were due to species or protocol differences. The results of these acute tests were negative or equivocal. Because the subchronic studies produced positive results, it was hypothesized that these carbodiimides required multiple treatments over an extended period of time to produce an increase in MN-erythrocytes. To confirm the original response, a second dermal subchronic study was conducted with DIC; the protocol was modified to include sequential blood samplings to permit monitoring MN frequencies over time. The data demonstrated a small but consistent induction of micronucleated erythrocytes in mice treated with DIC by skin painting.