The association between an area-based measure of deprivation and survival from the 10 most common cancers was studied in 155,682 patients diagnosed between 1980 and 1989 in the area covered by the South Thames Regional Health Authority. Furthermore, the impact of stage of disease at diagnosis on this association was studied. The measure of deprivation was the Carstairs Index of the census enumeration district of each patient's residence at diagnosis (5 categories) and the cancers studied were: lung, breast, colorectum, bladder, prostate, stomach, pancreas, ovary, uterus and cervix. In the univariate analyses the measure of outcome was the relative survival rate and in the multivariate analyses it was the hazard ratio. Both univariate and multivariate analyses showed that patients from affluent areas had better survival than patients from deprived areas for cancers of the lung, breast, colorectum, bladder, prostate, uterus and cervix. Stage of disease at diagnosis did not explain the survival differences by deprivation category. For cancers of the stomach, pancreas and ovary, no variation in survival by deprivation category was found. For most cancer sites, a clear gradient in survival by deprivation category was observed, which implies a large potential reduction of cancer mortality among the lower socioeconomic groups. Future studies need to incorporate other possible explanatory factors, besides stage, of the association between deprivation and survival.