INTRODUCTION: Cardiovascular disease (CVD) is usually caused by high levels of many risk factors simultaneously over many years. Therefore, it is of great interest to study if subjects stay within rank order over time in both the biological risk factors and the behaviour that influences these risk factors. Many studies have described stability (tracking) in single risk factors, especially in children where hard endpoints are lacking, but few have analysed tracking in clustered risk. METHODS: Two examinations were conducted 8 years apart. The first time, 133 males and 172 females were 16-19 years of age. Eight years later, 98 males and 137 females participated. They were each time ranked into quartiles by sex in four CVD risk factors all related to the metabolic syndrome. Risk factors were the ratio between total cholesterol and HDL, triglyceride, systolic BP and body fat. The upper quartile was defined as being at risk, and if a subject had two or more risk factors, he/she was defined as a case (15-20 % of the subjects). Odds ratios (OR) for being a case was calculated between quartiles of fitness in both cross-sectional studies. The stability of combined risk was calculated as the OR between cases and non-cases at the first examination to be a case at the second examination. RESULTS: ORs for having two or more risk factors between quartiles of fitness were 3.1, 3.8 and 4.9 for quartiles two to four, respectively. At the second examination, OR were 0.7, 3.5 and 4.9, respectively. The probability for "a case" at the first examination to be "a case" at the second was 6.0. CONCLUSIONS: The relationship between an exposure like physical fitness and CVD risk factors is much stronger when clustering of risk factors are analysed compared to the relationship to single risk factors. The stability over time in multiple risk factors analysed together is strong. This relationship should be seen in the light of moderate or weak tracking of single risk factors, and is strong evidence for early intervention in children where risk factors cluster.