OBJECTIVE: A new measure of body image, named the body image assessment for obesity (BIA-O) was developed and tested for reliability and validity in a sample of 1209 adult men and women. Separate BIA-O procedures were developed for men and women. Current, ideal and reasonable body image estimates of Caucasian and AfricanAmerican men and women were compared. METHOD: Figural stimuli of males and females were developed for body sizes ranging from very thin to very obese in 18 increments. Participants selected ®gures that represented estimates of current, ideal and reasonable (a body size that could be maintained over time) body size. Some participants (n 641) also completed two measures of body dissatisfaction in a test of the validity of the BIA-O as a measure of body dissatisfaction. A sample of 77 participants was administered the BIA-O on two occasions to test the test ± retest reliability of the BIA-O. RESULTS: The reliability of the BIA-O was supported by test ± retest reliability coef®cients which ranged from 0.65 to 0.93. Concurrent validity of the discrepancy between current and ideal and current and reasonable body size estimates was supported by positive correlations with two measures of body dissatisfaction. The BIA-O body size estimates of Caucasians and African-Americans, controlled for age and BMI, were compared. As BMI increased, Caucasian men and women were found to select larger current body size estimates in comparison to African-Americans. DISCUSSION: The reliability and validity of the BIA-O were supported. Greater body size dissatisfaction in obese Caucasians, relative to African-Americans of the same size, may be a function of biased estimates of current body size.