Patients with fragility fractures have low bone mineral density (BMD)--this statement is supported mainly by data on women. In this study, including only men, the objectives were to determine whether a decline in BMD alone or in combination with data on male sex hormones and skinfold thickness could be of value in predicting forthcoming fractures. We also wanted to find out whether high consumers of alcohol can be identified by measuring BMDs and male sex hormones. A prospective, population-based study was performed in the city of Malmö, Sweden. 242 men were randomly selected; all were of Scandinavian ethnic background, and were aged 50, 60, 70, and 80 years. Forearm BMD, testosterone, sex-hormone-binding globulin (SHBG), and skinfold thickness were analyzed. In addition, alcohol consumption and carbohydrate-deficient transferrin (CDT)--a marker of alcohol abuse--were analyzed. The study group was followed prospectively for 7 years and all fractures sustained were recorded. Prospectively, for a 1 SD decrease in forearm BMD, the Cox proportional hazard model gave a relative risk (RR) of 1.75 with a 95% confidence interval of 1.08-2.83 for a forthcoming fracture and 3.88 (1.30-11.57) for a hip fracture. For a 1 SD change in skinfold thickness, measured on the dorsum of the hand, a RR of 1.69 (0.99-2.87) for a forthcoming fracture was found and the corresponding value for hip fracture was 2.34 (1.10-5.00). Testosterone and SHBG did not enhance fracture prediction. Abusers of alcohol had, retrospectively, significantly more fractures. Individuals with alcohol consumption rates in the highest quartile had significantly higher CDT levels, but we were unable to identify high consumers of alcohol by analyzing BMD or sex hormones. In this study we found that forearm BMD and skinfold thickness could be used in predicting forthcoming fractures in men.