A method for collecting cancer-incidence data in a rural town is described. A mail survey, prompted and conducted by concerned citizens, was utilized. When analyzed in light of local mortality data and national cancer incidence and mortality rates, the data collected in such a survey seem reasonably accurate and complete. The numbers of observed and expected incident cases, as well as the numbers of observed and expected deaths, correlated quite closely for most sites. Excesses of incident cases were observed for seven of 11 sex-specific sites, but only for melanoma and lymphoma among women were the excesses statistically significant. Provided that community support is substantial, which is necessary to maximize accuracy and completeness, this method of collecting cancer-incidence data may be useful in rural areas of states without a statewide registry.