Evidence relating smoking and neuroticism is inconsistent; some studies report no significant relationship, while others report a positive association. The present study examined smoking in relation to neuroticism (N), extraversion (E) and psychoticism (P), in the light of predictions that interactions between N and E, and between N and P, could underly these inconsistent findings. The subjects were female student nurses (N = 270), of whom 34.1% were smokers. Discriminant analyses revealed that E differentiated most strongly between smokers and non-smokers. N and P did not add to the predictive power of the discriminant function, but the N x P and N x E interactions did make significant contributions. Among extraverts, and among subjects low in psychoticism, N was not related to smoking; but among introverts, and among subjects high in psychoticism, the relationship was significant and positive.