A retrospective cohort study, designed to examine the relationship between multiple chest fluoroscopies and subsequent breast cancer risk, was conducted in which 93.6% of 1764 former sanatorium patients were successfully located. The methodology used to trace former female patients who were treated for pulmonary tuberculosis between 1930 and 1954 (20-45 years before the study began) is presented. Extensive use of city directories and telephone resulted in locating the majority of study subjects. Outpatient medical records and state vital statistics departments were also of value. Searching for death certificates of parents was helpful as the informant named on the certificate was often the woman being traced or her husband. The Postal Service was of limited use. No difference in the proportion of breast cancer cases was found between the more easily traced patients and the last 9.7% of patients who were difficult to locate. Fewer of the last patients located had died, however, than those located without intense effort. Patients lost to follow-up were younger at the time of sanatorium admission, had a higher proportion of Protestants and non-white women, had shorter lengths of hospitalization and had less severe tuberculosis requiring less aggressive treatment than patients located.