We examined whether and how the provision of previous radiologic information (previous films and reports) influences the interpretation of radiographs. We prospectively studied 35 radiologists' interpretation of 311 plain-film radiology cases in a clinical setting. A radiologist first interpreted the current radiograph with only the information given on the consultation request. Subsequently, the same radiologist received, in random order, either the previous radiographs or the previous written reports, reviewed the diagnosis, changing it when necessary, noted the recognition of new findings, and adjusted his or her degree of confidence. A third interpretation used whichever type of information was not supplied for the second. All three readings of a study were performed at the same sitting by the same radiologist. A diagnosis and degree of confidence were recorded for each reading. The additional information, either radiograph or written report, significantly increased the confidence of the radiologist at each stage of interpretation. The largest increase in confidence occurred whenever previous films were introduced. In group A (second reading, reports; third reading, old films), new observations were made in 17.3% of cases on the second reading and in 19.9% on the third reading. In this group, diagnoses were changed in 14% on the second reading and in 9% on the third reading. In group B (second reading, old films; third reading, old reports), new observations were made in 16.9% of cases on the second reading and 7.3% on the third reading. Diagnoses were changed in 11% on the second reading and in an additional 5% on the third reading. Most changes were toward a more specific diagnosis. Prior information significantly increased the radiologists' confidence, facilitated new observations, and allowed more specific diagnoses. Prior radiographs were more valuable than reports in some respects.