The relationship between non-steroidal anti-inflammatory drugs (NSAIDs) and gastrointestinal events has been examined in Tayside by record linkage. From March to October 1983, 57 715 prescriptions for five NSAIDs (ibuprofen, indomethacin, naproxen, Osmosin and piroxicam) were issued to 25 959 patients. Their inpatient morbidity from 1 January 1983 to 31 December 1985 was compared to that of an equal number of control subjects matched for age, sex and general practitioner. Analysis revealed that NSAID takers were more likely to develop disease of the gastrointestinal system (ICD 520-579) and symptoms involving the gastrointestinal system (ICD 787). For females the relative risk was 1.67 and for males 1.28. The difference between NSAID takers and controls increased with age above 60 years and was more pronounced over 70 years. Perforation and haemorrhage of peptic ulcer and gastrointestinal haemorrhage was also higher among NSAID takers after prescription: this difference was dependent upon age but not related to sex. The morbidity of the two groups, control and subsequent NSAID takers, was comparable before the NSAID prescriptions were issued and the control rates remained constant throughout the study.