In a randomized, double-blind, placebo-controlled trial, 331 infants aged 6 to 12 months received orally, at an interval of 1 month, either two doses of live attenuated bovine rotavirus vaccine strain RIT 4237 or equivalent placebo. The vaccinations were carried out during September to November, a non-rotavirus season; only three cases of rotavirus diarrhea occurred in the study group before the vaccinations were completed. During the epidemic season from December to May, 31 patients with clinically significant rotavirus diarrhea required therapy. Five of these were among the 168 vaccine recipients, and 26 among the 160 placebo recipients (P less than 0.001), giving a vaccine protection rate of 82%. The incidence of clinically significant diarrhea from all causes was reduced by 76% in the vaccinees. As determined by an enzyme immunoassay antibody test with homologous virus antigen, seroconversion after vaccination was obtained in 53% of the initially seronegative infants. Clinical protection correlated well with seroconversion, but the vaccinees who failed to seroconvert also had less rotavirus diarrhea than the placebo recipients, suggesting that immunity may be mediated by factors other than serum EIA antibody. Seventeen of the 23 rotavirus isolates in the epidemic season that were typed were of serotype 1, two were of serotype 2, and four were of serotype 3. The protection rates against clinically significant diarrhea were 72%, 100%, and 100% for serotypes 1, 2, and 3, respectively. We conclude that epidemic infantile winter diarrhea associated with human rotaviruses can be significantly reduced by vaccination with the live attenuated RIT 4237 bovine rotavirus vaccine before the epidemic season.