The purpose of this study was to investigate the influence of physical activity and other factors on the mood of former elite male athletes and controls of middle and old age. The subjects were 664 former athletes and 500 controls who answered questionnaires in 1985 and 1995. The dependent variables depressive and anxiety symptoms were assessed by the shortened anxiety and depression scales of the BSI-53. Logistic regression was used for longitudinal as well as cross-sectional analyses to estimate odds ratios for symptoms of depression and anxiety in relation to leisure physical activity adjusted for age in 1995, sports group, personality characteristics, alcohol use, smoking, marital status, life events and socio-economic status. In the longitudinal analysis, low levels of physical activity as well as neuroticism, dissatisfaction, marital status, life events and social class in 1985 increased the risk of depression in 1995. Also physical activity has a protective effect against depressiveness; an increase of one MET-unit (hour/day) statistically significantly decreased the risk of depressiveness by 8 %. In the longitudinal analysis, physical activity had no significant association with anxiety. Cross-sectional analysis for depressive symptoms in 1995, but not for anxiety found associations with sports group and physical activity as well as alcohol use and marital status. Very high physical activity has a significant protective effect against depression.