Abstract Even though Eysenck’s N scales appear to be psychometrically superior to the E, P and L scales and compare favourably to similar scales of other questionnaires, they have been criticised for different reasons. The present research uses item response theory (IRT) to assess the psychometric properties of a pool of 47 N items drawn from different Eysenck questionnaires, and analyses criticisms based on nonequivalence, lack of unidimensionality, weak structure and social desirability bias. The results indicate that the pool of items is essentially unidimensional. The remaining criticisms were found to have some foundation but they did not invalidate the scales. The results of the IRT calibration provide further information about the properties of the items as measures of N.