The lipid results of a cross-sectional study of cardiovascular disease risk factors in over 600 13-18 year olds drawn from general practices in contrasting areas are reported. Differences were noted between the 3 practices, including a significant variation in the blood group distribution. HDL cholesterol showed a marked variation between the 3 practices which was unlikely to be due solely to seasonal variation. Alcohol consumption though not an univariate correlate of HDL cholesterol was however a strong partial correlate, as were age, height, sexual maturation, adiposity and smoking. Using 6 factors in each sex, 13% of the interchild variance of HDL cholesterol in boys could be explained in a multiple linear regression analysis, and 8% in girls. Including the practice variation as a further independent variable increased the explanation in 18% in boys and 15% in girls. An intraclass correlation (sib-sib) of 0.36 was found for HDL cholesterol, suggesting a strong familial influence. The implications of these results are discussed, along with the difficulties of screening for hyperlipidaemia in this age group.