Interest in computers is becoming commonplace in general practice. Many practices in the UK have already installed complete systems, others are in the processes of developing them. Most general practitioners will buy complete systems off-the-shelf, but a small minority will do their own systems design and programming. The computer, as part of an information system for general practice, will go through three phases: the cross-indexing system, the narrative record, and the decision aid. All systems currently installed are at the level of the cross-indexing system. This retains the traditional clinical record as the main repository of information stored in hard copy, but provides convenient and entropy proof methods of identifying cohorts of patients by such characteristics as age, sex, social class, geographical distribution, or disease entity. The purposes for this will be preventive medicine, quality control, administrative planning, educational planning, and possibly research. These functions replace manual systems which have been in use by a small number of general practitioners for twenty years, but which have never gained general acceptance partly because of their clumsiness and partly because they were seen as research tools. The step to the narrative record will not be worth the investment in resource or energy if it is purely to store the current undisciplined paper record in electronic medium. It will only be worthwhile if two preconditions are met: firstly, that some standardized form of recording such as problem-orientated medical records are accepted and, secondly, if it is accepted by the doctor that he is handing over a certain amount of control to the system which will prompt, cue and remind him of things that have to be done.(ABSTRACT TRUNCATED AT 250 WORDS)