Summary Data from the Cancer Research Campaign trial for early breast cancer have been used to study the effect of social class and weight on prognosis after primary treatment either by a simple mastectomy plus post-operative radiotherapy or by a simple mastectomy followed by a watch policy. There were 2455 patients for whom both social class could be determined and weight was recorded. These patients presented in clinical stages and 11 and were recruited between June 1970 and April 1975. The cut-off date for the analysis was 31 December 1991. When the survival curves of patients in manual classes were compared with those in non-manual classes, there was a tendency for the latter to do better, but the difference was not statistically significant (P = 0.12). By contrast, there was a highly significant difference (P = 0.002) in survival favouring patients weighing less than or equal to 60 kg compared with those weighing greater than 60 kg. The difference was confined to post-menopausal patients and was still highly significant when included in a multivariate analysis with social class, age, tumour size, clinical stage and tumour grade. The effect of weight was to increase the mortality due to breast cancer rather than other causes.Keywords: breast cancer mortality; social class; weightThe effect of social class on overall mortality is well recognized (Blane et al, 1990), manual workers tending to have higher rates than non-manual. Most studies of the effect of socioeconomic factors on survival after treatment for breast cancer have suggested a similar trend, namely a worse prognosis in lower socioeconomic classes. A recent review by Schrijvers and Mackenbach (1994) of six studies found in four of them a statistically significant increased relative risk (RR) of dying for patients with the lowest socioeconomic status. A paper published subsequent to this review also found a clear gradient by deprivation category, with better survival for women from more affluent areas (Schrijvers et al, 1995).There have also been numerous studies on the effect of weight on prognosis in patients treated for breast cancer. Goodwin and Boyd (1990) reviewed 14 such studies before 1990 and identified a modest effect of body size on prognosis (smaller women doing better), the effect being greatest in post-menopausal women and in those with little or no involvement of axillary nodes. Nevertheless, some more recent reports (Ewertz et al, 1991;Gordon et al, 1992;Katoh et al, 1994) have been unable to show convincingly an improvement of prognosis with decreasing weight or some index of obesity.These studies vary considerably in patient numbers and in whether or not adjustments have been made for other prognostic factors. In addition, their conclusions have often been based on follow-up periods of 5 years or less, so that the considerable number of deaths from breast cancer that occur more than 5 years after first treatment have not always played a part in the analyses.