The serum high density lipoprotein (HDL) subfractions, HDL2, and HDL3, and serum apolipoprotein AI and B (apo AI and B) were evaluated as potential indicators of the risk of ischaemic heart disease in men aged less than 60 years who had previously had a myocardial infarction and in controls with a similar socioeconomic background who had no history of myocardial ischaemia. Discriminant analysis confirmed that the combination of serum cholesterol, triglycerides, and total HDL cholesterol distinguished poorly between patients and controls. The best single discriminating variable was apo B. Stepwise discriminant analysis showed that this discrimination could be improved to a small extent by combining apo B with apo AI and parental history, but nothing was gained by measurement of serum cholesterol triglycerides, very low density lipoprotein cholesterol, low density lipoprotein cholesterol, HDL cholesterol, HDL2 or HDL3 cholesterol. Significantly more patients than controls with type IV hyperlipoproteinaemia had raised concentrations of serum apolipoprotein B, but the frequency of raised apolipoprotein B concentrations was no greater in patients with type IV hyperlipoproteinaemia than in those with normal serum lipids. The value of apo B as an indicator of cardiovascular risk should be assessed in prospective studies.