The 1991 Peruvian cholera epidemic has thus far been responsible for 600,000 cholera cases in Peru. In an attempt to design a cholera surveillance program in the capital city of Lima, weekly sewage samples were collected between August 1993 and May 1996 and examined for the presence of Vibrio cholerae O1 bacteria and V. cholerae O1 bacteriophages (i.e., vibriophages). During the 144 weeks of surveillance, 6,323 cases of clinically defined cholera were recorded in Lima. We arbitrarily defined an outbreak as five or more reported cases of cholera in a week. The odds of having an outbreak were 7.6 times greater when V. cholerae O1 was present in sewage water during the four previous weeks compared with when it was not (P < 0.001). Furthermore, the odds of having an outbreak increased as the number of V. cholerae O1 isolations during the previous 4 weeks increased (P < 0.001). The odds of having an outbreak were 2.4 times greater when vibriophages were present in sewage water during the four previous weeks compared with when they were not, but this increase was not statistically significant (P = 0.15). The odds of having an outbreak increased as the number of vibriophage isolations during the previous 4 weeks increased (P < 0.05). The signaling of a potential cholera outbreak 1 month in advance may be a valuable tool for the implementation of preventive measures. In Peru, active surveillance for V. cholerae O1 and possibly vibriophages in sewage water appears to be a feasible and effective means of predicting an outbreak of cholera.