The article by Lodwick, Keats, and Dorst (1), appearing in this issue of Radiology, raises some urgent questions concerning the use of computers in medicine. The future demands answers. We must decide if computers can serve medicine as they have begun to serve other fields of science. And, while we struggle with that first question, we must simultaneously contend with another—one that troubles not only physicians, or even scientists, but is a prime problem of our time—that is, how much can man rely upon these electronic assistants of his which are at once super- and subhuman? The answer to the first question is, I believe, an assured “yes.” It has become almost impossible to handle the large quantities of knowledge which are necessary to the proper management of disease without the aid of some mechanical or electronic device. Computers are now being developed which can prove theorems, play games with sufficient skill to beat their inventors, translate texts from one language to another, recognize the spoken word, and learn to improve their own performance. Considering the significance of probabilities in medical diagnosis and management, it is obvious that any procedure which can maximize the wins and minimize the losses will be of great value. By means of electronics we can determine probabilities, from complex data, far more rapidly and accurately than could be done by any individual or group. Computers are particularly effective when handling objective problems which end themselves to mathematical analysis. An example of this can be seen in the use of these machines to calculate the results of chemical examinations of the blood. By feeding a binary logic computer the laboratory determined values of only four or five blood chemicals, one can predict accurately the values of some thirty others without resorting to further laboratory procedures. Although far less objective than chemical determinations, roentgen signs, as opposed to symptoms and physical findings, do lend themselves to digital designation. For this reason the computer principle should be of particular interest to radiologists. It is well therefore, that the article by Lodwick and his associates so vividly demonstrates the value of computers in radiological research. What they have proposed is one of the simplest applications of the computer; namely, an acceleration of the process of turning clinical data into research data. Essentially, the computer as used in this study is simply a data calculator, combining findings which have been converted into numbers and giving answers on a probability curve as to the significance of these signs for prognosis. Such answers, of course, could be obtained by investigators using simple calculating machines, but the enterprise would require an enormous number of hours of work and is therefore unlikely to be undertaken.