The effect of the duration of retinoid treatment on the inhibition of 1-methyl-1-nitrosourea-induced mammary carcinogenesis was studied. Female Sprague-Dawley rats were given i.v. injections of 50 mg 1-methyl-1-nitrosourea per kg body weight at both 50 and 57 days of age. Feeding of a placebo diet or diet supplemented with 323 mg retinyl acetate per kg diet (retinoid treatment) was initiated at 10 days after the first carcinogen injection. Retinoid treatment was either continued or discontinued after 60 days postcarcinogen, and the study was terminated at 182 days postcarcinogen. Retinoid treatment between 10 and 60 days postcarcinogen prolonged the cancer latency and reduced the average number of cancers per rat in comparison to that in placebo-treated rats. Continuation or cessation of retinoid treatment in 60-day tumor-bearing rats had no effect on the time of appearance of additional cancers. In 60-day tumor-free rats, continuation of retinoid treatment prolonged cancer latency in comparison to either 60-day tumor-free rats changed to placebo or rats continuously treated with placebo. The cessation of retinoid treatment in 60-day tumor-free rats resulted in a rapid increase in the appearance of cancers; at the termination of the study, the average number of cancers per rat was similar to that of animals fed only the placebo. The data indicated that some rats are more responsive to the retinoid than are others. Retinoid treatment apparently prevented the progression of early neoplastic lesions, and a continuous daily intake of the retinoid appears necessary to sustain the chemopreventive effect under the experimental conditions imposed.