A 98.8% response rate was achieved in a follow-up study of a community based health survey. The follow-up methodology used in contacting the residents of 1700 households in a suburban upper-middle class community is presented. Study subjects, who were predominantly white, older adults, participated in the original health survey during 1972 and 1973. When the follow-up study was conducted three years later to determine the vital status of study participants, 30% of the households had moved. Follow-up response of two sets of mailings totaled 76.1%. An additional 12.1% response was obtained by calling residents at their last known address. The final 10.5% of the contacted households were located by intensive telephone tracing. A wide variety of sources of information, both local and non-local, were used to locate these "hard to trace" households. They were different from all other contacted households with respect to mobility rate, household size, and death rate. A 15.5% underestimate of the three-year death rate would have occurred if the hard to trace group had not been contacted. These results emphasize the importance of complete follow-up in prospective studies.