The research objective of this work is development of statistical methodology for identifying 'non-tracking' individuals through their longitudinal observations. Non-tracking individuals are those whose patterns of change over time differs qualitatively from that of most other individuals in the population Following methods developed by Rao, patterns of change over time are modeled as linear combinations of optimally chosen orthogonal vectors. Non-trackers are defined as individuals requiring more than the usual number of vectors to fit their set of observations. The Akaike Information Criterion is used to determine the number of vectors required by each individual, and hence to identify the nontrackers. The statistical methodology developed is applied to simulated data to study the reliability of the method. It is also applied to longitudinal observations of cardiovascular disease risk factor variables measured in children. Height, systolic blood pressure and serum total cholesterol were analyzed. In general, children 5-17 years of age track for these variables. The number of vectors fitted for each age cohort showed the methodology correctly modeled important biologic events such as the height growth spurt in children