Colonic tumorigenesis involves the processes of initiation and promotion/progression from normal epithelial cells to tumors. Studies in both humans and experimental models of colon cancer indicate that secondary bile acids promote tumor development. In contrast, we have demonstrated previously that another bile acid, ursodeoxycholic acid (UDCA), inhibits the development of azoxymethane (AOM)-induced colon cancer in rats. More recently, we have shown that UDCA inhibits AOM-induced hyperproliferation, and aberrant crypt formation and growth. In our previous studies, we supplemented UDCA throughout the experiment. The efficacy of a chemopreventive agent may depend on the timing of administration, which has important clinical implications. In the present investigation, we examined the ability of UDCA, when administered only in the initiation or the promotion/progression phase, to block tumor development. Male Fisher 344 rats were divided in a 2 x 3 factorial design, with animals receiving AOM or vehicle, and fed an unsupplemented diet or diet supplemented with 0.4% UDCA in the initiation or promotion/progression phase. Thirty-two weeks later, rats were sacrificed and tumor histology determined, and colons were examined for aberrant crypt foci (ACF). In the carcinogen-treated dietary control group, tumor incidence was 72.3%, and tumor multiplicity was 1.9 tumors per tumor-bearing rat. UDCA, in the initiation or promotion/progression phase, significantly decreased tumor incidence to 46.2% and 38.4% (P < 0.05), respectively; and tumor multiplicity to 1.4 and 1.3 tumors per tumor-bearing rat (P < 0.05), respectively. UDCA did not alter tumor size, histology, or location, although there were trends for smaller tumors and less advanced histological grades in the group given UDCA during the promotion phase. UDCA, in the initiation but not the promotion phase, inhibited ACF formation and growth. In summary, UDCA significantly inhibited AOM-induced colonic carcinogenesis during either tumor initiation or in the promotion/progression phase. In contrast, UDCA inhibited ACF formation only when administered in the initiation phase, suggesting that the mechanisms of chemoprevention by this bile acid differ in these two phases.