The full range of premature mortality and associated risk factors was analyzed for a follow-up period of three and a half to eight years in a uniform group of 7,935 middle-aged males (46 to 48 years old at screening) participating between the years 1975 and 1979 in the preventive population program in Malmö (participation rate 76.7 percent). Of the 218 deaths that occurred, necropsy was performed in 181 (83.0 percent). Three major causes of death were established: cancer (61/218), alcohol-related deaths (55/218), and coronary heart disease (50/218). In these three main categories of male premature mortality, significant and distinctly differential risk factor patterns were found. In coronary heart disease, smoking (p = 0.0062), serum cholesterol level (p = 0.00014), serum triglyceride level (p = 0.00013), systolic blood pressure (p = 0.000012), and diastolic blood pressure (p = 0.0021) were the strongest single determinants, but the independent role of the diastolic blood pressure disappeared in a multivariate analysis whereas all the others could be combined in a highly predictive logistic model. In the alcohol-related group, equal or stronger risk factor associations were present for serum gamma-glutamyltransferase level (p less than 0.0001), questionnaire alcoholism screening response (p less than 0.0001) and, inversely, serum cholesterol level (p = 0.0046) and serum creatinine level (p less than 0.0001), all of which were independent and could be combined in an even more predictive logistic model than in the coronary heart disease group. In the cancer deaths, significant associations were found for serum urate level (p = 0.023) and, inversely, serum cholesterol level (p = 0.056 - 0.031). Malignant and alcohol-related diseases constituted at least equally prominent groups as the cardiovascular disorders of the total premature deaths that occurred during middle age in these cohorts of Malmö males. All of these conditions are potentially avoidable and seem to be associated with significant and distinctive risk factor patterns. It seems possible that these factors may be applied, in current alcohol-related disorders and in future malignant diseases, both as indicators of the respective risks and as signals and instruments for directed preventive measures like the previously well established and tested methods for the regulation of blood pressure, serum lipids levels, and so on.