Social differences in health concern both ethics and science. From a public health point-of-view, one must assess actual differences and then try to find explanations. This was made possible for the first time for cancer in Italy via nationwide record-linkage between the 1981 census and the national death index. Over the subsequent six months after census, the study-base included 31,000 deaths for cancer and 18 million person-years at risk. Rate ratio (RR) were estimated through a Poisson regression model adjusted by age and geographic area of residence. Educational level was used as social level indicator. Profound social differences were evident for buccal cavity (RR = 3.10 for lowest cf highest educational level), esophagus (RR = 3.00), stomach (RR = 3.43), and larynx (RR = 3.30) among men, and for stomach (RR = 2.25) and uterus (RR = 1.76) among women. Colon (RR = 0.62) and pancreas cancers (RR = 0.65) presented an inverse relationship among men, as did colon (RR = 0.37), breast (RR = 0.56), ovary (RR = 0.45), and melanoma (RR = 0.62) among women. In conclusion, the Italian population at the beginning of the 1980s had large social differences in the risk of dying from cancer, confirming the patterns commonly found in such other countries as Great Britain, France, and New Zealand. Some dissimilarities, useful for hypothesis generation on the mechanisms of inequality, were evident, such as the generally highest social differences found among northern Italian men and among southern Italian women.