It was reported previously that a score composed of glucose, uric acid, and triglyceride was correlated closely with the total cholesterol:high density lipoprotein cholesterol (TC:HDL) ratio and could be used to predict those subjects for whom the more expensive HDL measurement would be useful. The authors tested this score using multiple regression and discriminant function analysis in a population sample of 209 men aged 40-79 years, Triglyceride was the only component of the score that contributed to the prediction or classification of the TC:HDL ratio. Although the correlations of the score with the TC:HDL ratio were relatively good (r = 0.72 or greater), the discriminant function analysis showed that subjects with high or low scores could have high or low TC:HDL ratios. The authors conclude that the score (essentially the triglyceride level) is inadequate to permit the classification of any individual as having a definite high or low TC:HDL ratio.