In the two most recent cholesterol-lowering drug trials, the achieved reductions in coronary heart disease mortality were offset by increases in mortality due to accidents and violence. A possible biochemical explanation has been suggested for an association between low serum cholesterol level and increased risk of death due to injury. We, therefore, examined the association between serum cholesterol level and risk of death from accidents or violence in the 25-year follow-up of two cohorts of Finnish men (N = 1580). Although a statistically nonsignificant, negative association was observed in one cohort (hazard ratio, 0.84, with a 1 mmol/L increase in cholesterol), the other cohort showed a statistically significant, positive association in multivariate analysis (hazard ratio, 1.39). We conclude that the observed associations between serum cholesterol and deaths from injury in the present study and in cholesterol-lowering trials are probably determined by other, presently unknown factors, or by chance.