There are conflicting reports on the effect of soy and its components on mammary carcinogenesis in adult female rats, mainly because of different rodent models that are used in chemoprevention studies. The present study was undertaken to compare the tumor-preventative effects of soy protein isolate (SPI) and two of its isoflavones in a "standard" model that had been used for the identification of many chemopreventive agents. Six groups of female Sprague-Dawley rats were provided with modified cornstarch AIN-76A diets supplemented as follows: no additional agents (control), purified genistein (200 mg/kg diet), purified daidzein (200 mg/kg diet), genistein + daidzein (100 mg/kg diet each), SPI containing normal levels of isoflavones (SPI-n), or SPI depleted of isoflavones (SPI-d). Mammary carcinomas were induced by 7,12-dimethylbenz[a]anthracene (DMBA) introduced 1 wk after the animals began consuming the experimental diets. At the end of the study (120 days after DMBA treatment), no significant differences were found among the six groups with respect to tumor incidence or survival, nor was there a significant reduction in tumor multiplicity in the genistein or genistein + daidzein group. However, there was a 32% reduction in tumor multiplicity in the daidzein and SPI-n groups relative to the control group (P < 0.05). The most effective diet was SPI-d, which produced a 50% reduction in tumor multiplicity relative to the control (P < 0.01). The difference between the SPI-d group and the daidzein or SPI-n group was not significant. Median tumor latency was increased from 53 days in the control group to 68 days in the daidzein group and to 72 days in the SPI-d group, but these differences were not statistically significant. These results show that daidzein and SPI (with normal or low levels of isoflavones) are effective inhibitors of DMBA-induced mammary tumors in adult rats.