The measurement of high-density lipoprotein (HDL) for the purpose of assessing the risk of cardiovascular disease in the individual subject was evaluated. Three laboratory methods were compared, two electrophoretic and one heparin-manganese precipitation, and the HDL results were expressed both as a percent and as an absolute concentration. In phase 1 of the study, the optimal method and the best cut point were identified. In phase 2, these were applied to a larger population who were assigned, on the basis of clinical criteria, to a coronary heart disease and to a control group. The overall probability of correct classification of an individual by his HDL result was calculated. When HDL was expressed as a percent and determined by gel electrophoresis, 82.6% of control subjects and 83.0% of patients with coronary heart disease were classified correctly using the optimal cut point of 23.5%.