A systematic review of randomized clinical trials was conducted to evaluate the acceptability and usefulness of computerized patient education interventions. The Columbia Registry, MEDLINE, Health, BIOSIS, and CINAHL bibliographic databases were searched. Selection was based on the following criteria: (1) randomized controlled clinical trials, (2) educational patient-computer interaction, and (3) effect measured on the process or outcome of care. Twenty-two studies met the selection criteria. Of these, 13 (59%) used instructional programs for educational intervention. Five studies (22.7%) tested information support networks, and four (18%) evaluated systems for health assessment and history-taking. The most frequently targeted clinical application area was diabetes mellitus (n = 7). All studies, except one on the treatment of alcoholism, reported positive results for interactive educational intervention. All diabetes education studies, in particular, reported decreased blood glucose levels among patients exposed to this intervention. Computerized educational interventions can lead to improved health status in several major areas of care, and appear not to be a substitute for, but a valuable supplement to, face-to-face time with physicians.