We used a high-level adaptation paradigm to distinguish between two hypotheses: (1) perceptual and attitudinal body image measurements reflect conceptually different mechanisms which are statistically independent of each other; (2) attitudinal (e.g., questionnaire) and perceptual (e.g., visual yes-no) body image tasks represent two different ways of measuring exactly the same construct. Forty women, with no history of eating disorders, carried out the experiment. Each participant carried out five adaptation blocks, with adapting stimuli representing female bodies at: extreme-low body mass index (BMI), mid-low BMI, actual BMI of the observer, mid-high BMI, and extreme-high BMI. Block order was randomized across participants. The main outcome variable was percentage error in participants’ self-estimates of body size, measured post-adaption. In regressions of this percentage error on the strength of the adapting stimuli together with observers’ attitudinal body image as a covariate, we found positive regression slopes and no evidence for any interaction between the fixed effects. Therefore, we conclude that perceptual and attitudinal body image mechanisms are indeed independent of each other. In the light of this evidence, we discuss how people with eating disorders, like anorexia nervosa, may come to over-estimate their body size.