This study reports the validity, reliability, and responsiveness of a 33-item fat- and fiber-related behavior questionnaire (FFB) and describes how this instrument provides insight into the process of adopting healthy diets. Data are from the Eating Patterns Study, a randomized clinical trial of a physician-delivered, self-help intervention to reduce fat and increase fiber intake. Intervention (n = 850) and control participants (n = 945) completed both a food frequency questionnaire and the FFB at baseline and at 3 and 12 months postintervention. Validity, as assessed by correlation of the FFB with the food frequency questionnaire at baseline, was 0.53 for fat (fat scale with percentage energy from fat) and 0.50 for fiber (fiber scale with fiber g/1000 kcal; both P < 0.001). Reliability, as assessed by the intraclass correlation in controls across all three time points, was 0.77 for the fat scale and 0.74 for the fiber scale (both P < 0.001). The largest changes in fat-related behavior were in avoiding fat as a flavoring and in using specially manufactured low-fat foods, and the largest changes in fiber-related behavior were in substituting high-fiber versions of common foods. Overall, the FFB was a reasonably valid and reliable measure of dietary intake, which provided insight into the behavioral effects of the dietary intervention.