Two pilot tests were made of computer-aided self-help for chronic agora/claustrophobia diagnosed by a psychiatrist. Test 1 was of 17 patients at the Institute of Psychiatry (IoP). They had no human clinical help, getting ideas only from the computer on how to do self-exposure therapy and set exposure goals week by week. Patients could attend for 12 sessions. Outcome data were inconsistent across ratings, not always matching clinical impression of progress. Of the 15 patients for whom outcome data were available, 6 improved markedly or moderately on clinical impression. Test 2 was of 6 patients in a rural general practice (GP) in Wales who were given help and support from a nonclinical administrator. Of the 4 Welsh patients with follow-up data, 3 improved markedly or moderately. The test results suggest desirable changes. Patients first need to do practice ratings before rating on the computer and to check computer ratings. Some rating instructions need to be clarified. Most importantly, synergism comes from giving patients access to brief human help if they encounter problems when using computer-aided self-help.