Background. We sought to develop a refined measure of eight personality traits or constructs observed in those who develop depression. We report the psychometric properties of the derived Temperament and Personality (T&P) questionnaire, as well as a pilot study examining its capacity to differentiate over-represented personality traits in those with depression.

Method. The factor structure of the T&P measure was examined in a general practice sample of 529 subjects. We imposed a range of factorial solutions to determine how higher-order molar constructs arborized to eight lower-order constructs. Scale scores generated at each derived tier were contrasted for 52 out-patients with major depression and control subjects from the general practice sample to pursue over-represented personality constructs, and to clarify if an optimal number of constructs could be identified.

Results. In the factor analysis, some 90% of the items loaded on their a priori construct. The questionnaire showed high internal consistency, test–retest reliability and minimal sensitivity to mood state effects. Analyses rejected the hypothesis that risk to depression might be generally affected by individuals merely scoring high on all ‘normal’ personality styles, whether higher-order or lower-order traits.

Conclusions. Findings suggest that, while identified constructs linked well with the widely accepted theoretical model of personality (the Five Factor Model) at one tier, such a fixed model may be too inflexible. We therefore detail potential advantages to using a multi-tiered model of personality traits in application studies.