Repertory grid testing (Kelly, 1955; Slater, 1965) of psychiatric patients, using people of significance in the patient's life as elements and eliciting the patient's own constructs has been employed in the clinical setting to demonstrate aspects of psychopathology (Ryle, 1967; Ryle, 1969), to measure change with treatment (Ryle and Lunghi, 1969; Rowe, 1971a) and to assess how far a therapist could predict a patient's responses (Rowe, 1971b). However, no systematic examination has been reported so far into the characteristic features of the construct systems of neurotic as opposed to normal subjects, although some apparent equivalences between grid features and psychological or psychopathological formulations such as identification or splitting have been described. While repertory grid testing remains an essentially ideographic exercise, it is important to identify features characteristic of neurosis if the method is to have more than a descriptive function in the investigation of patients. As a contribution to this we have tested with an identical method a series of students consulting with neurotic problems and a control sample of new students tested on arrival at the University.