As women with breast hypertrophy may have their breasts surgically reduced, the natural question is whether this operation influences their risk for breast cancer. Animal experiments have suggested that the damage to the mammary ducts caused by mammoplasty leads to the evolution of stasis and hence to cancer (Fekete and Green, 1936). It has also been suggested, however, that surgical reduction of the breast decreases the incidence of breast cancer by reducing the number of potential foci for cancer development (Str6mbeck, 1964;Rees and Coburn, 1972).This study is a second follow-up of a cohort originally studied by Lund et al. (1987), comprising 1283 women operated on for breast hypertrophy between 1 January 1943 and 31 August 1971, who were followed up to 31 December 1982. The original study found a reduced incidence of breast cancer among these women in comparison with the Danish female population. The cohort has now been followed up to 31 December 1990. Materials and methodsThe study group consisted of all women with breast hypertrophy treated by reduction mammoplasty at five surgical departments in Copenhagen, Denmark, between 1 January 1943 and 31 August 1971, and three patients operated on before 1 January 1943. The material was collected by examining diagnostic indices and lists of operations. The hospital records of 31 patients could not be located and these patients were excluded from the analysis. In the first study five patients were included twice in the study population, which therefore consisted of 1278 women. Six patients were lost to follow-up; in our follow-up, one patient was excluded as she appeared to have died before the operation. Thus, our study group consisted of 1240 patients. Table I).We also looked at the risk of breast cancer on the basis of how much breast tissue was removed from the breast with the greatest reduction: <400 g, 400-600 g or >600 g. The 367 women with more than 600 g of tissue removed had a significantly lower risk than the general population (RR=0.30; 95% CI 0.10-0.69) at all time intervals after operation (latency). In addition, Table I shows how the RR varies with latency time. There appears to be a sharp drop in the RR 10-19 years after operation and an increase