Linked record medical information systems generate cumulative person or family records from data on events or processes occurring to the individual at separate times and places. The special purposes of time-based and family statistics from these systems are exemplified by studies in health care, epidemiology and genetics. Methods of linking records on a large scale and some of the difficulties are described. The value and practicability of computer-assisted linked record systems are discussed in the light of experience with the Oxford Record Linkage Study and the northeast Scottish Psychiatric Case Register, both of which have been in existence for a decade.