An excellent and concise summary of the rationale for the intermittent concern some physicians have shown about the possible assistance computers may render the diagnostician was recently advanced by Bruce and Yarnall.1 Whenever he makes a decision on diagnosis, prognosis, or choice of therapy, the physician faces the possibility of being "wrong." Can the computer help minimize the number and seriousness of such errors? Allegations and claims concerning the ability of computers to function as diagnosticians have accompanied the growth of medical computer applications from the beginning. Sufficient experience with computers has accumulated by now so that we can examine the possibility of computer-based diagnosis more critically. Such a careful evaluation is also called for because many astute and thoughtful physicians have been repelled by exaggerated and misinformed claims of what the computer can and cannot do. These exaggerations detract from the real service computers can eventually and indeed