All 67-year-old pensioners in a primary care district (N = 142) participating in a multi-disciplinary population study were followed until the age of 83. At 83 years of age, 65 persons had survived and continued to take part in the study. Social, psychological and medical factors predicting survival during the period have previously been reported (Samuelsson et al. 1992). In the present analysis, the same variables at age 67 were used to predict health, measured with six different health indicators, at 83 years. The analysis has been performed separately for women and men. Variables at 67 years of age as determinants for health at age 83 have been identified and ranked through successive selection in a step-wise discriminatory analysis. For women, reported diseases at 67 was a very strong predictor but quite the contrary for men. Blood pressure and sleep medication were strong predictors for men but not for women. Psychological factors were more frequently included in the predictive models for women than for men. Social factors were of comparatively less importance for both men and women. There was greater accuracy in the prediction of health for women. The individual variables most often included in the predictive model for women were coping and reported health at age 67. For men, blood pressure, sleep medication and intelligence were the most frequent predictors. The analysis demonstrated clear sex-specific prediction patterns. When comparing predictors for survival and predictors for differentiated health in the same population no similarities were found.