Two methods of long-term epidemiological follow-up were compared by using each to study the survival of 1622 myocardial infarction patients registered by the Belfast MONICA Project. Length of follow-up ranged between 3 and 5 years during which time 277 deaths were recorded. A computer-based method for linking MONICA Project registration records with the Registrar General's death certification data identified 273 of the 277 deaths. Follow-up supplied by the Northern Ireland Central Services Agency through the flagging of patients in their master patient index identified 271 deaths; four of the six deaths which were missed occurred before computerisation of the index was complete. The study illustrates the value of computer-based linkage with death certification data and of flagging in the Central Services Agency master patient index.