The authors analyzed tracking and predictiveness of serum lipid and lipoprotein measurements in Finnish children and young adults over a 12-year follow-up period. A representative sample of 3,596 healthy subjects aged 3-18 years was examined in 1980. The follow-up studies were done in 1983, 1986, 1989, and 1992. Data were available on serum lipids and lipoproteins, anthropometric measurements, dietary and smoking habits, and use of oral contraceptives. Complete data on serum lipids in 1980 and 1992 were available for 883 subjects (47% males), and they comprised the study cohort for this analysis. Significant tracking was found in each of the serum lipid variables studied. The range of 12-year correlations was 0.48-0.58, 0.53-0.58, 0.53-0.58, 0.57-0.59, and 0.33-0.37 for serum total cholesterol, low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, the LDL:HDL cholesterol ratio, and triglycerides, respectively. Males showed more tracking than females; there was no clear age trend. Tracking of HDL2 cholesterol was better than that of HDL3 cholesterol (0.64 vs. 0.43, respectively; 3-year tracking). Apolipoproteins A-I and B showed similar amounts of tracking compared with HDL and LDL cholesterol, respectively. Approximately 50% of subjects who initially fell into the extreme quintiles of total cholesterol, LDL cholesterol, and HDL cholesterol were in the same quintiles after 12 years. In multiple regression analyses, childhood obesity, exercise, diet, and smoking habits did not markedly aid the prediction of adult serum lipid values. However, the use of two childhood measurements increased the amount of adult serum lipid variability explained. Although universal screening cannot be endorsed, these findings emphasize the importance of serum lipid measurements in the early detection of familial lipoprotein disorders and in the initial evaluation of coronary heart disease risk in childhood.