Objectives: To study natural weight changes and to develop a weight classification system that can identify weight maintainers, gainers, and losers. Design/outcome: A prospective, observational study in which weight changes over five annual measurements were evaluated. In the weight classification system used, changes greater than 5 lb defined weight maintenance, gain, or loss. Subjects/settings: Subjects were healthy, normal- weight and overweight, men and women (mean age=44.1±14.1 years) in the Relationships of Energy, Nutrition, and Obesity to Cardiovascular Disease Risk Study. Prospective data for 385 of the original 508 subjects for whom actual weights were available for each of the 5 years (1985 to 1990) were used to classify and characterize subjects by weight-change categories. Statistical analyses: Cross-tabulations (with χ2tests) and hierarchical log-linear analyses (with partial χ2tests) to examine the relationships of categorical variables; analyses of variance (with F tests) for continuous measures. Results: Over the 4-year interval, 46% of subjects were classified as maintainers, 34% as gainers, and 20% as losers. Over shorter 1-year epochs, more subjects were maintainers (62%) and fewer subjects were gainers (22%) or losers (16%). Maintainers had fewer and smaller magnitudes of weight fluctuations and showed fewer deleterious changes in health risk factors than gainers. Applications: Weight changes of greater than ±5 lb can classify a person as a weight maintainer, gainer, or loser. Although annual weight changes were used in this study, a weight change of more than 5 lb between any two points in time may suggest nonmaintenance of weight or weight instability that needs further evaluation.