In this practice, with a family practitioner committee list of 9726 patients, we use a computer register for recall, screening, morbidity data, audit, and repeat prescribing. The computing techniques used to achieve accuracy in maintaining the register are described. After one year of full use the register was validated by using the computer to select a random sample of 200 patients from patients' computer records that had not been updated recently. Two patients were untraceable, and in only 11 records were errors of information found, none of which was important. We think that it is feasible and valuable to have a household index.