A major problem in the analysis of attrition of cohorts in studies on mental health problems is that data on those who do not participate at the outset of a study are largely unavailable. It is not known how underlying psychopathology affects the fi rst stages of screening where non-response and selectivity are usually highest. This article presents results of one of the centres of the Netherlands Study of Depression and Anxiety (NESDA), a longitudinal study aimed at describing the long-term course and consequences of depression and anxiety disorders. The aim is to describe the different ways of attrition during the fi rst NESDA-wave in a cohort of patients aged 18-65 years of the Registration Network Groningen and to analyse whether attrition is related to gender, age and psychopathology as recorded in general practice. The attrition of the study cohort (n = 8475) was highest during the fi rst stages, eventually leading to a population of 169 patients only who participated in the full NESDA-programme. Probabilities of transition from one stage of the screening process to the next were regressed on selected background variables using binary logistic regression. Correlates of participation were being female and being older (>40). Psychopathology was an important variable in the formation of the initial sample cohort, but only had a weak infl uence on patient response to the screening questionnaire. Study design factors had a stronger impact on the changing composition of the cohort at each screening stage compared to patient factors.