In past decades, there were many surveys on adolescents problem behavior (in terms of problem behavior theory formulated by R. Jessor and colleagues). Self-report questionnaires are widespread method for data obtaining in such studies. Our paper points out some validity problems that arise with relying on self-report data such touchy, as the items in questionnaires related to problem behavior are. Providing an information about own problem behavior can be perceived as very menacing for a respondent, no matter how much discreetness could researcher promise. In addition, specific social groups appreciate or directly demand some forms of behavior, which is undesirable by the norms of conventional society and thus labeled as problem. These pressures can result in distorted responses about problem behavior. Responses on items can be magnifying, denying, or there can be unusual number of missing values. With exploratory analysis performed on data obtained in two large surveys realized in Czech Republic (California Project, European Longitudinal Study of Pregnancy and Childhood; total N = 1200) from 1996 to the present day, we identified factor of items related to severe forms of problem behavior, which could be seen as a score of specific social desirability as well. On the second hand, group of respondents with numerous missing values showed significantly greater scores on correctly answered items. These results challenge the validity of self-report questionnaires targeted to problem behavior.