OBJECTIVE--Research into the health of minority ethnic groups is often restricted by methodological difficulties. These include the lack of accurate population denominators, the choice of an appropriate sampling frame, correctly assigning ethnic group, and biases in techniques used for sampling and investigation. This article reviews the available sources or mortality and morbidity data, and assesses their uses and limitations for research involving ethnic minority groups. Suitable sampling frames and review methods used to assign ethnicity are discussed. Sources of bias are high-lighted and methods used to overcome these biases are presented. CRITERIA FOR INCLUSION OF ARTICLES--Articles have been chosen which best illustrate the problems encountered and show how these problems can be addressed. CONCLUSIONS--The increased documentation of ethnic origin on routine data sources is welcomed, but attention must be paid to ensuring that congruent definitions in data collection are used. The worrying consequences of the Commission of European Communities directive, which describes the need for explicit consent to be obtained from subjects before data is used for anything other than its original purpose, are discussed.