In computer-assisted telephone interviewing, questions are displayed on a computer screen, and responses are entered directly into a computerized data file. In 1981-1982, a randomized trial of computer-assisted telephone interviewing, compared with telephone interviewing with responses directly recorded on printed questionnaires, was carried out. The respondents were surrogates for 400 white Florida residents who died in 1979 and were randomly selected from a death certificate-based case-control study of colorectal cancer. Outcomes examined included participation rate after initial phone contact, length of interview, recorded number of comments, recorded number of probes, unresolved "don't know" responses, and the interviewer's evaluation of the quality of the interview. The computer-assisted telephone interviewing system resulted in the 25-30-minute interviews lasting, on the average, 3.4 minutes (14%) longer. The average number of comments decreased from 5.5 to 4.1 (a 25% difference) and probes from 10.2 to 8.3 (a 19% difference) in the computer-assisted interviews. These differences were markedly smaller than the differences noted between individual interviewers.