To determine the relation between breast self-examination and survival of breast cancer patients, the authors studied 1004 newly diagnosed invasive breast cancers in Vermont general hospitals from July 1,1975 to December 31, 1982. Physicians were asked to determine whether patients had been performing breast self-examination 1) at least monthly, 2) at some interval less than monthly, or 3) never. Breast self-examination data were obtained on 83 per cent of the patients. For the analysis in this report, noninvasive forms of breast cancer (N = 35) have been excluded from the data set. All pathology reports and any operative reports were examined. The alive/dead status of patients was determined at least annually. Deaths were coded as due to breast cancer or other than breast cancer based on the opinion of the treating physician at the time of death or by review of the autopsy report and/or the death certificate. Five per cent of surviving patients had last follow-up dated prior to 1982. Two per cent were definitely lost to follow-up data. Of the 1004 patients, 836 had known breast self-examination performance prior to detection of breast cancer. Of these, 23 per cent reported they had been performing breast self-examination at least monthly; 28 per cent, less frequently than monthly; and 49 per cent, never. Of the 697 patients for whom family history was known, 10 per cent had a family history of breast cancer in a first-degree female relative (sister and/or mother). Patients performing breast self-examination more frequently were likely to have a family history of breast cancer. There was an association between more frequent performance of breast self-examination prior to detection of breast cancer and earlier clinical and pathological stages of breast cancer at discovery. Women performing breast self-examination more frequently were significantly more likely to detect their breast cancer themselves. The time period between first recognition of a symptom or sign of breast cancer and the histologic diagnosis of breast cancer was somewhat shorter in women performing breast self-examination. There was little relationship between the length of delay and breast self-examination frequency when monthly breast self-examination performers were compared to less than monthly examiners. At the time of this analysis, the median follow-up time for surviving patients was 52 months with a maximum follow-up of 92 months. There were 242 deaths among the 835 invasive breast cancer patients with both breast self-examination performance data and follow-up data. Breast cancer was determined to be the cause for 70 per cent of the deaths. The authors collapsed the monthly and less than monthly breast self-examination performance groups into a single group (performers) because of the limited number of deaths to date in these two groups. At the time of this analysis, there has not been a significant difference in survival between the monthly breast selfexamination group and the less than monthly breast selfexamination group. The product limit survival curves both for all deaths and deaths due to breast cancer were significantly better (P < 0.001 by log-rank test) for breast self-examination performers than for nonperformers. At 5 years, the overall survival of breast self-examination performers was 75 per cent (+3 per cent) versus 57 per cent (+3 per cent) for nonperformers. Deaths due to breast cancer had occurred in 14 per cent of the breast selfexamination performers compared to 25 per cent of the nonperformers (P < 0.001 based on x2). The survival difference persisted after adjusting in the log-rank analysis for the covariate's age, family history of breast cancer, delay in treatment, and method of detection when taken singly or in any of the possible combinations for both breast cancer deaths and all causes of death (all P < 0.002). When the patients were subdivided by age, there were significant differences of similar magnitude in survival within each of the age groups, with breast self-examination performers being less likely to die of breast cancer in each age group. Examination of the monthly death rates indicates that a large part of the difference in overall survival between breast self-examination performers and nonperformers is due to the differences in deaths due to breast cancer.