Lipoprotein(a) [Lp(a)] has been considered to be a predictor of premature coronary heart disease and other cardiovascular diseases. Lp(a) levels are largely genetically determined, but the detailed mechanism of Lp(a) elevation is uncertain. We examined the association between Lp(a) levels and apolipoprotein(a) [apo(a)] phenotypes as well as that of Lp(a) level and other various conditions. The subjects were 280 healthy Japanese (102 males and 178 females) aged 39 to 70 years who were living in a rural community in 1992. We obtained apo(a) phenotypes determined by SDS-PAGE as well as Lp(a) levels and other cardiovascular risk factors. We combined apo(a) phenotypes form 4 groups according to molecular weights (from high apo(a) molecular weight to low: I, II, III and IV). Lp(a) levels were associated with apo(a) phenotype-groups, that is, they were inversely associated with apo(a) molecular weight. Small apo(a) phenotypes were less frequent than large ones. The median Lp(a) level was higher in smoking (29.2 mg/dL) than in non-smoking subjects (18.5 mg/dL) in phenotype-group III. Adjusted means of total cholesterol and fibrinogen levels in apo(a) phenotype-group IV were the highest of all phenotype-groups. Age, apo(a) phenotype, smoking status, total cholesterol and fibrinogen were positively correlated with Lp(a) levels by multiple regression analysis. Lp(a) levels were found to be mainly associated with apo(a) phenotype, but varied broadly within the same apo(a) phenotype at various conditions, such as smoking status and high total cholesterol.