Heavy users of the services of emergency departments (EDs) have in previous studies been found to have psychological, social, economic and other difficulties besides their more or less acute medical problems. In order to establish whether mortality is associated with high ED use, a nine-year follow-up study was conducted of a 10 per cent population sample (n = 17,000), selected from the catchment area of Huddinge Hospital, Sweden. ED visits were found to predict nine-year mortality in the cohort. The group of individuals who had made four or more ED visits during a period of 15 months prior to follow-up (heavy ED users) had a two-fold excess mortality (95 per cent confidence interval (CI) = 1.9-2.1), those who had made one to three ED visits (moderate ED users) had a slightly elevated mortality (standardized mortality ratio SMR = 1.1, 95 per cent CI = 1.0-1.3), while the SMR of the non-users was 0.9 (95 per cent CI = 0.8-1.0). The three predominant causes of death in the cohort were diseases in the circulatory system, tumours and violent death. Heavy ED users had elevated mortality in all diagnoses, the most important excess mortality being from violent death, comprising suicide, probable suicide and alcohol/drug abuse, with an SMR of 6.3 (95 per cent CI = 6.0-6.7). The excess mortality from these causes of the heavy ED users accounted for more than one-third of their total excess mortality.(ABSTRACT TRUNCATED AT 250 WORDS)