On-study barium radiographs of 535 patients in the National Cooperative Crohn's Disease Study have been analyzed for the pattern of distribution of bowel disease and the individual features that characterized bowel involvement. On-study and off-study radiographs of 403 of these patients were compared under code to judge radiographic response to drug treatment and discover correlations of radiographic findings with clinical response. Patients with more clinically active disease had more colonic disease on x-ray. Duodenal abnormalities were recognized in 22% of the patients and radiographically typicaly Crohn's disease of the duodenum in 8%. Recurrent Crohn's disease and that characterized by small bowel obstruction each displayed a characteristic appearance. Overall there was little evidence of radiographic improvement during the study, and little correlation between clinical response and evidence of radiologic improvement. Only patients treated with prednisone for more than 6 mo showed statistically significant radiologic improvement. Patients with definite radiographic progression or regression were found in each treatment group. Both fistula and stricture with obstruction were associated with a poor clinical response to all therapies. In view of the evidence from this study that radiographic findings do not correlate with clinical symptoms or response, the ritual use of x-ray to follow patients with Crohn's disease is unnecessary.