West European societies have become increasingly culturally diverse. In this article we discuss two issues that arise when conducting criminological research in diverse contexts, namely threats to study validity as a result of (a) ethnic differences in participation rates and (b) lack of cross-cultural equivalence of questionnaires in multi-language contexts. We demonstrate how these problems were handled in the z-proso study, a prospective longitudinal criminological study involving 1240 families in Zurich, Switzerland. Results suggest that recruitment efforts can significantly increase participation rates, particularly among minority members. As expected, higher participation rates are associated with a qualitative improvement in the sample: parents who needed to be personally motivated to participate tend to have children with higher rates of problem behaviour than do parents who participated spontaneously. Highest rates of problem behaviour were found for those children whose parents did not participate. Finally, the multi-method analysis reveals significant discrepancies in parent rating of child problem behaviour among migrant parents compared with teacher and interviewer rating. Closer scrutiny indicates that such differences are likely to be a result of socially desirable response behaviour among low-education parents.