We examine the psychometric validity of the Ambiguous Intentions Hostility Questionnaire (AIHQ) by applying confirmatory factor analysis (CFA). Previous analysis of psychometric features of the AIHQ showed problems with internal consistency and stability of the tool over time. By using CFA we wanted to check whether the scenes depicted in the questionnaire actually measure the same construct and whether the subscale measures are intercorrelated. The analysis included ambivalent scenes and was conducted on offenders (N = 108) and the general population (N = 161). Results indicated that the structure proposed by the AIHQ authors differs from that revealed by the statistical procedures. It appears that the AIHQ factor structure is centered on situational contexts rather than on particular questions measuring constructs of hostility and blame. We discuss these findings with regard to application of the AIHQ in clinical samples.