The phenomenon of item nonresponse, i.e. missing data, in surveys is well known among methodologists. Item nonresponse is a problem when it is biased to the dependent variables in aetiological research. The occurrence of item nonresponse in self-reported delinquency studies has been associated with the threatening nature of questions about previous delinquent behaviour, but item nonresponse also occurs in scales measuring aetiological variables (theoretical concepts) in aetiological research, and in sociology has also been associated with negative attitudes towards the survey, although evidence from self-reported delinquency studies in support of this concern has not yet been given. The aim of this study is to evaluate the seriousness of the problem of item nonresponse in two independently drawn self-reported delinquency data sets of two classroom delinquency studies conducted among adolescents in Antwerp (Belgium) and Halmstad (Sweden) using paper and pencil interviews (PAPI). We do this by evaluating the non-random character of item nonresponse in scales of delinquency and aetiological variables, by looking at the correlates of item nonresponse and by evaluating the effects of assigning values on the missing data with regard to reliability and correlational validity. The results are rather optimistic about the hypothesized negative effects of item nonresponse.