Data from an outpatient treatment trial for anorexia nervosa were examined to gain preliminary insights as to whether the modified Stroop colour-naming task might offer a useful measure of treatment outcome. It was hypothesised that interference for eating-, weight- and shape-related words on a modified version on the Stroop colour-naming task would decrease with successful treatment. Studies to date suggest that eating disorder patients show interference on the Stroop colour-naming task for food- and shape-related words. The study was designed to assess selective-information changes for patients with anorexia nervosa before and after treatment in association with other psychological measures. Pretreatment and post-treatment assessments were collected from a clinical trial of 25 women who met "Diagnostic and Statistical Manual of Mental Disorders" (4th ed., "DSM-IV") criteria for anorexia nervosa. Using repeated measures analysis of variance, a significant Time x Card interaction was found on the Food and Control Food Stroop cards, indicating that interference decreased following treatment. The results of this preliminary study raise numerous questions for further research. These findings suggest that the modified Stroop colour-naming task may be sensitive to change among anorexia nervosa patients and may potentially provide a valuable measure of treatment outcome. Comparison with a nonclinical group needs to be made to determine whether the reduction in Stroop scores is a normative process. Furthermore, studies with larger sample sizes are needed to compare treatment responders and nonresponders with matched controls to rule out the possible influence of a practice effect. (Contains 2 tables.)