The hypothesis that recurrent or chronic depressive illness produces a long-term change in neuroticism was examined in a sample (N = 34) from a consecutive series of 89 depressed patients admitted to the Maudsley Hospital in 1965/6. The Eysenck Personality Inventory (EPI) was administered at the time of the index illness both when the patients were depressed and on recovery, and then again at follow-up 18 years later. The change in the neuroticism (N) score over the 18-year-period was compared in good and poor outcome groups defined variously by a global rating of outcome, frequency of episodes, extent of subsequent hospitalization and the presence or absence of subsequent chronicity. The mean N score for the sample as a whole did not change significantly over the 18 years, and no differential change in the N score was observed between any of the good and poor outcome groups. Thus, the hypothesis was not supported.