Objective: To investigate the extent to which changes in physical activity predict changes in abdominal fat in women over an 18-month period, while statistically controlling the effects of possible confounders, such as age, total body fat percent, and energy intake. Design: A prospective cohort design over 18 months. There was no intervention or treatment. Changes in objectively-measured physical activity were used to predict changes in abdominal fat over the study period. Subjects: 110 healthy, middle-aged women (mean: 41.3±3.3 yrs), primarily Caucasian, educated, and married. Measurements: An objective measure of physical activity (ACT) using CSA accelerometers, worn continuously for 7 consecutive days at baseline and again at follow-up. Total body fat and abdominal fat percent were assessed by dual energy x-ray absorptiometry (DEXA). Energy intake was estimated using 7-day, weighed food records for the days in which subjects wore accelerometers. Results: No significant change between baseline and follow-up means for abdominal fat, physical activity, or energy intake over the study period. Moreover, change in physical activity was not a significant predictor of change in abdominal fat, with or without statistical control of confounders. Change in energy intake was a predictor of abdominal fat (P=0.0688), and this association was strengthened after adjusting for age, baseline total body fat, and changes in physical activity. Conclusions: Apparently, when measured using accelerometers, changes in physical activity are not predictive of changes in abdominal fat over an 18-month period. However, changes in energy intake seem to predict changes in abdominal fat. Evidently, increases and decreases in abdominal fat are more a function of energy intake than physical activity in middle-aged women.