to the results of annual healtl examinations over 4 years. In this study, 2,511 mal subjects aged 40-69 yr were asked to fill in th, questionnaire. Of these, 824 subjects had a healtl examination in 1993. Two hundred and twenty-eigh of these remained within the normal range in live function tests (LFT) in each of annual healtl examinations between 1994 and 1996, whereas 63 hay worsened LFT results after the initial examination and were therefore called the decreased liver function (DLF group. Controls age-matched to the DLF group were chosen from the normal liver function group Discriminant function analysis was carried out to identif the differences between the DLF group and the contrc group. The independent variables were selected fron the variables which showed significant differences ii the univariate analysis, considering multicollinearity The independent variables used for the discriminan analysis were therefore alcohol consumption, diastolic blood pressure (DBP), BMI and "aggression" by THI The Wilks' Lambda value was 0.787 (p value=0.0001) The overall correct identification rate in the discriminan analysis was 75.0%. The results of the present stud suggest that men who are heavy drinkers hypertensive, obese and aggressive had a high risk c developing liver dysfunction within a few years, even i their LFT values were within normal limits in any singly year. (J Occup Health 1998; 40: 339-344)