Summary Objectives: Despite evidence of poor health outcomes associ- ated with excessive weight gain or loss, longitudinal patterns of body weight over the adult life course have not been fully described. This article seeks to address this by examining body weight patterns for middle-aged and older adults. Methods: Panel data from the Health and Retirement Study (HRS) original cohort and the original cohort of the Asset and Health Dynamics Among the Oldest Old (AHEAD) study are used to compare, by social group, characteristics of respon- dents and non-respondents, baseline weight (1993 and 1994) to year 2000 weight, and explore various weight change tra- jectories over time. Results: Overall, a greater proportion of middle-aged adults are heavier over time than at baseline and a greater proportion of older adults lose weight over time compared to baseline. Ex- amining the transitions across weight trajectories for all of the social groups suggests considerable variability. Conclusions: These findings suggest that differing physiologic and social or environmental experiences may have implications for body weight patterns. Understanding these patterns by race/ethnicity and gender could provide insight into health dis- parities among different social groups. Keywords: BMI – Obesity – Longitudinal analysis – Weight change trajectories. The specter of a worldwide epidemic of obesity has alarmed health care providers and policy makers in developed and developing nations alike (World Health Organization 1998). Obesity is of such concern because of its associations with poor health outcomes for individuals and substantial eco- nomic burdens to society (McIntyre 1998). In the United States, where a major public health initiative seeks to raise the proportion of normal weight adults from 42 % to 60 % (U.S. Department of Health and Human Services 2000), the percentage of the population that is overweight or obese has increased dramatically in recent decades (National Heart Lung and Blood Institute 1998). Although the aging of the American population alone cannot explain this finding, data from the National Health and Nutrition Examination Sur- veys (NHANES) and the Behavioral Risk Factor Surveil- lance System (BRFSS) do suggest that body mass index (BMI) increases throughout the adult life span until early old age (Mokdad et al. 1999; Galuska et al. 1996; Flegal 2000; Flegal et al. 1998). This indicates that the second half of life may present an important focus for obesity research. The NHANES and BRFSS programs utilize repeated cross- sectional studies to describe weight trends. While providing important information, these aggregate time series data can mask off-setting individual-level changes. To investigate the latter requires longitudinal data in which the same panel of respondents is observed or interviewed repeatedly over time. There is little documentation of longitudinal patterns of body weight for American adults even though obesity is a profound issue in the U.S., and particularly so for minority groups in which obesity is known to be high (Lee et al. 2001). As Lee et al. (2001) note, this gap in knowledge is unfortu- nate: “Understanding patterns of weight change over time may allow us to identify critical periods for emphasizing pre- vention of weight gain, because we have little success cur- rently with long-term maintenance of weight loss” (Lee et al. 2001). Given the strong link between overall health and health risk, such an understanding may help target public Soz.- Präventivmed. 48 ( 2003) 257 – 268 0303-8408/03/040257–12 DOI 10.1007/s00038-003-2053-3 © Birkhäuser Verlag, Basel, 2003 Kristi R. Jenkins 1 , Nancy H. Fultz 1 , Stephanie J. Fonda 2 , Linda A. Wray 1, 3, 4 1 Institute for Social Research, The University of Michigan, Ann Arbor 2 Joslin Diabetes Center, Harvard Medical School, Boston 3 University of Michigan Medical School, Ann Arbor 4 Pennsylvania State University, University Park Patterns of body weight in middle-aged and older Americans, by gender and race, 1993 – 2000 Section: International comparison of health determinants