Objective: The number of elderly migrants from Turkey and Morocco in Western Europe will increase sharply in the coming decades. Many of these migrants seem to have mental health problems. As mental health care programs are based on DSM criteria, there is a need for diagnostic instruments with good psychometric properties. This exploratory study examines the presence of construct bias, method bias and item bias in the Composite International Diagnostic Interview (CIDI) Basis Life time version 2.1 in elderly Turkish and Moroccan men and women. Method: From a community based health survey four groups of migrants aged 55-74 were selected for semi-structured interviews including the CIDI depression section. Data included interview transcriptions and observations of 11 respondents in each group (Turkish men, Turkish women, Moroccan men, and Moroccan women). The data were analysed using qualitative techniques. Results: Construct bias, method bias and item bias of the CIDI was found in all groups. The poor match between the CIDI on the one hand and the taboo on mental health problems and the poor level of education of the respondents can partly explain this bias. Conclusions: The use of the CIDI in elderly migrants of Moroccan and Turkish descent is problematic, due to the presence of construct, method and item bias. Copyright © 2005 John Wiley & Sons, Ltd.