Data were obtained for all deaths registered between 1979-1983, and for all new cancers recorded at the Victorian Cancer Registry between 1982-1983, in residents of Melbourne. A socioeconomic status (SES) measure had been produced for each local government area (LGA) by principal components analysis of sociodemographic variables recorded at the 1981 census. A SES score from 1 to 10 was assigned to each death and cancer. Population data from the census were similarly scored. Age standardised rates for all cause mortality, for mortality from all causes other than cancer and for both incidence and mortality of total cancers, cancer of the stomach, colon, rectum, lung, female breast, cervix, uterus, prostate and bladder, and for melanoma, lymphoma and leukaemia were analysed as a function of SES decile using weighted linear regression. Despite the limited number of years of data and the misclassification of the SES score, analyses showed there were inequitable distributions of mortality, and of some major cancers, across social strata in Melbourne during the early 1980s. The incidences of cancer of the breast, colon, prostate and melanoma were all positively associated with SES, while the incidences of cancer of the stomach, lung and cervix demonstrated negative SES gradients. For cancers where incidence showed a significant SES gradient there was a similar SES gradient with mortality. These patterns are consistent with the literature and implicate SES differences in education and access to services. Implications for health policy are discussed.