There is substantial evidence linking steroid hormones and diet to cancer aetiology. The evidence on lifestyle determinants of steroid hormones, however, is limited. We have conducted a study to identify dietary and other lifestyle predictors of testosterone and oestradiol among adult men. Subjects were 112 healthy Greek men, recruited as controls in a case-control study on the aetiology of liver cancer. Demographic data and detailed histories of smoking habits and alcohol consumption were recorded. Diet was assessed through an interviewer-administered validated food-frequency questionnaire. Serologic measurements of oestradiol, testosterone and sex hormone binding globulin were also conducted. We developed linear regression models to evaluate the associations of smoking and dietary factors with serum testosterone and oestradiol. The results indicate that, among men, both testosterone and oestradiol serum levels decline with age, whereas body mass index may be inversely related with testosterone and positively with oestradiol. The evidence concerning alcohol in relation to these hormones is inconclusive. Emerging evidence concerning smoking suggests positive associations with both hormones in the blood. The principal nutritional findings are a positive association of carbohydrate intake with testosterone levels and a set of inverse associations of vitamins with oestradiol.