Two hundred nine female enrollees of the Group Health Cooperative of Puget Sound who developed advanced-stage breast cancer during the period 1982-1988 were interviewed about their practice of breast self-examination (BSE), use of other breast cancer screening modalities, and medical and reproductive histories. Each subject's description of how she performed the examination was scored according to her mention of up to 10 recommended BSE techniques. A random sample of 433 women without advanced-stage breast cancer from the same population was interviewed for comparison. Relative to women not practicing BSE, the risk of advanced-stage breast cancer among BSE users was 1.15 (95% confidence interval, 0.73-1.81). Frequency of BSE did not differ between women with advanced-stage breast cancer and control subjects, whether in all subjects or in subgroups defined by age, use of mammography, or frequency of clinical breast examinations. While self-described proficiency in BSE was generally low in both case and control subjects, the small percentage of women reporting more thorough self-examinations, regardless of frequency, had about a 35% decrease in the occurrence of advanced-stage breast cancer compared to women who did not perform BSE. These results suggest that, while carefully performed BSE may avoid the development of some advanced-stage breast cancers, BSE as practiced by most Seattle-area women is of little or no benefit.