OBJECTIVE To determine whether serum sialic acid concentration may be used to predict short and long term cardiovascular mortality. METHODS Prospective study on all men and women who had their serum sialic acid concentration measured as part of a general health survey in 1964 or in 1965. All were followed up for an average of 20.5 years. METHODS Geographical part of the county of Värmland, Sweden. METHODS Residents in the area participating in a health check up in 1964-5 (27,065 men and 28,037 women), of whom 372 men (169 with incomplete data and 203 lost to follow up) and 345 women (143 and 202 respectively) were excluded; thus 26,693 men and 27,692 women entered the study. The study sample was restricted to subjects aged 40-74 during any of the 20 years' follow up. METHODS Serum sialic acid concentration, serum cholesterol concentration, diastolic blood pressure, body mass index at the general health survey visit; cardiovascular and non-cardiovascular deaths during three periods of follow up (0-6 years, 7-13 years, and 14-20 years), according to the Swedish mortality register, in subjects aged 45-74. RESULTS Mean serum sialic acid concentration (mg/100 ml) was 68.8 (SD 8.0) for men and 69.2 (8.0) for women; the average concentration increasing with age in both sexes. A total of 5639 (21%) men and 3307 (12%) women died during the follow up period, in whom death in 3052 (54%) men and 1368 (41%) women was from cardiovascular causes. During short (0-6 years), medium (7-13 years), and long (14-20 years) term follow up the relative risk of death from cardiovascular disease increased with increasing serum sialic acid concentration. The relative risk (95% confidence interval) associated with the highest quartile of sialic acid concentration compared with the lowest quartile was 2.38 (2.01 to 2.83) in men and 2.62 (1.93 to 3.57) in women. Similar results were found for deaths from non-cardiovascular disease with relative risks of 1.50 (1.34 to 2.68) in men and 1.89 (1.57 to 2.28) in women, but these relative risks were significantly lower than those for deaths from cardiovascular disease (p less than 0.001 and p less than 0.005 respectively). In multivariate analysis of total mortality and of cardiovascular mortality with sialic acid concentration, serum cholesterol concentration, diastolic blood pressure, and body mass index as independent variables the impact of sialic acid concentration was virtually the same as in univariate analysis. CONCLUSIONS Serum sialic acid concentration is a strong predictor of cardiovascular mortality. A possible explanation of these findings is that the serum sialic acid concentration may reflect the existence or the activity of an atherosclerotic process, and this may warrant further investigation.