This paper reports the results of methodological studies carried out in conjunction with the US National Comorbidity Survey (NCS) to evaluate Version 1.0 of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). These studies relied on recent survey data collection methodology literature to investigate problems regarding question comprehension, instruction comprehension, respondent motivation to report accurately, and regarding the limits of respondent ability to report accurately. Insights and strategies developed by survey methodologists were used to modify the CIDI in an effort to address these problems. The paper describes these strategies and methodological studies that evaluated their effects, including a clinical reappraisal study and a field experiment that evaluated the impact of question modifications on prevalence estimates. The paper closes with a discussion of remaining methodological problems with the CIDI and potentially useful future studies that might be able to develop solutions to these problems. Copyright © 1998 Whurr Publishers Ltd.