Studies of mortality from asthma and chronic obstructive pulmonary disease (COPD) have relied on death certification or registration for case finding. The aim of this study was to determine the accuracy of death certification and registration in asthma and COPD. All death certificates in Northern Ireland for 1987 where asthma or COPD (defined as International Classification of Diseases 9th Revision (ICD9) 490, 491, 492, 496) were listed in part I or part II were identified. The following certificates were then selected for further investigation: those mentioning asthma for all ages, those mentioning COPD for ages less than 56 years, and a 50% sample of those mentioning COPD aged 56-75 years. For these selected deaths the general practitioners' case notes, hospital records, and necropsy findings were reviewed. Questionnaires detailing the clinical history and circumstances of death were completed by the general practitioner by post and by a close relative or associate of the deceased (doctor administered) if, after initial investigation, the death was likely to be due to COPD or asthma. A panel of two respiratory physicians reviewed each death and, using clinical diagnostic criteria, assessed the accuracy of the registered cause of death. Of 50 registered asthma deaths 43 were confirmed as being due to asthma. In nine registered deaths from COPD in cases aged less than 56 years one was confirmed as COPD, two as asthma, and six as other respiratory conditions. Of 105 registered deaths from COPD in cases aged 56-75, 42 were confirmed as COPD, 27 as asthma, eight as other respiratory conditions, and 28 as other causes. Although few errors in registration were found, 21% of certificates mentioning asthma and 38% of certificates mentioning COPD but not asthma in part I were subject to variable application of the classification rules by the registering officers. For all deaths under 75 years of age in Northern Ireland in 1987 where either asthma or COPD was mentioned anywhere on the death certificate, the estimated sensitivity and specificity of the registered cause of death in predicting the "true" cause of death were 29% and 98.6% for asthma and 69% and 70% for COPD. In a population of subjects where asthma or COPD was mentioned anywhere on the death certificate, the registered cause of death is a relatively poor indicator of the "true" cause of death for both asthma and COPD. Variation occurred in the application of death classification rules by registration officers. Many deaths certified and registered as COPD could have been called asthma using current standards of clinical diagnosis. In studies investigating risk factors for deaths from asthma, case finding should consider deaths registered as COPD.