OBJECTIVE: Our goal was to examine 12-week covariability in diet and physical activity changes among 8- to 10-year-old African-American girls and if these changes predicted percent change in BMI. RESEARCH METHODS AND PROCEDURES: Covariability among percent changes [(post - pre)/pre x 100] in nutrients, food groups, and physical activity was assessed among 127 8- to 10-year-old African-American girls. Pearson correlation and hierarchical linear regression analyses were performed. RESULTS: Percent change in percentage kilocalories from carbohydrate was negatively correlated with percent change in both percentage kilocalories from fat (r = -0.85; p < or = 0.01) and protein (r = -0.51; p < or = 0.01). No statistically significant relationships were observed in percent changes among food group variables. Negative relationships were observed between percent changes in fruit/100% juice and percentage kilocalories from fat (r = -0.20; p < or = 0.05) and between percent changes in minutes of moderate-to-vigorous and sedentary activity (r = -0.60; p < or = 0.01). No significant associations were observed between percent change in moderate-to-vigorous physical activity and diet variables or percent change in BMI or waist circumference and percent change in diet or physical activity. DISCUSSION: No relationships were observed between percent changes in physical activity and dietary variables. Percent change in diet and/or physical activity did not predict percent change in BMI. This may have been due to the small sample size, the small changes in diet or physical activity, the short duration of the intervention, or because data from different interventions were combined. Understanding these relationships could have significant implications for addressing the obesity epidemic.