This paper describes the relationship between social deprivation and incidence of, and survival from, breast, lung, and colorectal cancers among residents of the South Thames regions. We analysed 23,505 cases of breast cancer, 29,903 cases of lung cancer and 21,905 cases of colorectal cancer, aged 40-99 inclusive at diagnosis and diagnosed between 1 January 1987 and 31 December 1992. Using the 1991 Census in conjunction with the Townsend index on social deprivation, we derived proxy indicators of deprivation based on patients' home postal codes. Cumulative relative five-year survival rates (per cent) were calculated for each cancer. We then compared our results with the relevant standardized incidence and mortality ratios by deprivation status. A clear trend was observed in standardized mortality rates across deprivation tenths for the three tumour sites: mortality increased with deprivation. A strong positive correlation was found between deprivation and the incidence of lung cancers (p < 0.0001), but no association was found between deprivation and incidence of breast and colorectal cancers. Significantly lower five-year relative survival rates were found for breast and colorectal cancer patients in the most deprived Townsend tenths. Breast cancer patients resident in the most affluent tenth of enumeration districts had a 70 per cent relative survival ratio compared with 57 per cent in the most deprived tenth. The corresponding figures for colorectal cancer patients were 40 per cent and 32 per cent, respectively. Survival differences by deprivation status exist in South Thames among patients suffering from breast or colorectal cancers and are not explained by differences in the incidences of these diseases. For lung cancer, incidence and mortality were positively correlated with deprivation, but no socio-economic gradient was found for survival.