Abstract
Computers, formerly regarded as research tools, are attaining increasingly wide clinical use in nuclear medicine. Physicians and technologists who wish to purchase computer systems are usually inundated with confusing and sometimes contradictory specifications by salesmen and manufacturers. Relatively few of these data, however, are important in determining whether a given computer can perform the functions for which it is required at a reasonable speed. These critical specifications are defined and their pertinence to the cost-effectiveness of various computer systems is discussed. Likely developments in computer design and construction during the next few years are also described.