A total of 368 workers from six paint-manufacturing factories participated in this study. The workers were classified according to type of exposure: direct, intermittent, and no exposure. The workers' liver-function tests were influenced greatly by gender, hepatitis B, alcohol consumption, and body mass index. Both the serum concentration and the odds of abnormality of total serum bile acids were elevated among the directly exposed group. The authors concluded that analysis of covariance should take into account occupational and nonoccupational factors on liver-function tests to avoid any errors. Total serum bile acids also indicated liver dysfunction from solvent exposure.