The reduction of socio-economic inequities in health is now an explicit objective of health policy in Canada. This study examines changes in mortality by income in urban Canada from 1971 to 1986 in terms of both relative and absolute differences between income groups. Street address information as shown on death certificates was used to code census tract of usual place of residence for deaths occurring to residents of Canada's Census Metropolitan Areas (CMAs) in 1971 and 1986. After exclusion of residents of health care institutions, 73,995 deaths were included in the study for 1971, and 88,129 for 1986. These deaths were analyzed by income quintile (based on census tract incidence of low income), age, sex, and cause of death. In 1971, the difference in life expectancy at birth between the highest and lowest income quintiles was 6.3 years for men and 2.8 years for women. By 1986, these differences had decreased to 5.6 years for men and 1.8 years for women. However, relative mortality (lowest compared to highest income quintile) at most ages changed only slightly over the 15 years. Relative infant mortality, for example, was 1.97 in 1971 and 1.82 in 1986. In 1986, 21% of total potential years of life lost (PYLL) prior to age 75 could be attributed to differences in quintile death rates compared to rates for the highest income quintile. Approximately 45% of this "excess" PYLL was for persons under 45 years of age. In 1971, the comparable figure was 67%. In 1986, the major causes of death contributing to income inequalities in mortality were: circulatory diseases, accounting for 25% of excess PYLL related to quintile differences; accidents, poisonings and violence, accounting for about 17%; and neoplasms, accounting for 15%. Respiratory diseases, ill-defined conditions, metabolic diseases and perinatal conditions each contributed 6-7% of excess PYLL. From 1971 to 1986, in terms of age-standardized morality rates (ASMRs) for all ages, certain causes of death showed increased mortality together with greater inequality by income, especially for males: these causes included lung cancer, suicide, metabolic diseases other than diabetes, and ill-defined conditions. Other causes of death showed either little change or less inequality by income but higher ASMRs: these included breast cancer, colon and rectal cancer, arterial diseases, alcoholism, mental disorders, and diseases of the nervous system.(ABSTRACT TRUNCATED AT 400 WORDS)