OBJECTIVE: To analyse the association between the Swedish underprivileged area (UPA) score and the standardized mortality ratio in Swedish municipalities. SETTING: All 284 municipalities in Sweden. DESIGN: The UPA-score was calculated for municipalities using the proportion of persons in the following groups: elderly persons living alone, children under five, persons in one-parent families, unskilled (SEI 1), unemployed, persons living in crowded households, those who have moved house in the last year, and persons of minority ethnic origin. After transformation (arc sin square root of) and standardization, each of the eight variables was weighted by the UK general practitioners' average weighting and added to give a composite index of socioeconomic deprivation--the UPA-score. The strength of the relationship between the UPA-score and premature mortality in Sweden was investigated by regression analysis using SMR, for people aged 20-64 years, 1989-93, as the dependent variable. RESULTS: The mean UPA-score (standard deviation) of the 284 municipalities was 0 (10.0) and the range -43 to 28. The mean SMR for all persons was 1 and the range 0.46 to 1.81. The association between UPA-score and SMR was statistically significant and the UPA-score explained 20% of the variation between municipalities in SMR. CONCLUSION: The range of social deprivation at the municipality level in Sweden, as measured by a composite index such as the UPA score, is wide. A two-fold variation at municipality level was also found in premature mortality. There was a significant association between high rates of mortality and social deprivation.