We studied tracking of serum lipids in children and young adults (n = 248) during a 12-year follow-up. A novel method was developed for assessing tracking. It was based on the calculation of the absolute width of the individual track consisting of 5 serum lipid determinations. The total cholesterol (TC) track width tended to be broader in females compared to males (1.33 versus 1.24 standard deviation (SD) units). Approximately 35, 39, 24 and 13% of subjects had all 5 measurements within 1 SD (of the cohort mean) from serum TC, low density lipoprotein cholesterol, high density lipoprotein cholesterol and triglycerides, respectively. However, variations in excess of 2.5 SDs were not rare. We also assessed the dependence of tracking on the absolute lipid level by calculating the SD of the standardized 6-year change in a cohort of 2236 children. In each serum lipid variable studied, lowest degree of tracking was found in highest lipid values. The mean changes were also largest in extreme values due to regression toward mean phenomenon. Our findings indicate that the width of longitudinal track of serum lipids may be two wide to reliably detect high risk children. Further, the track width was even broader in subjects with extreme serum lipid values.