Depressive illness is known to be associated with low self-evaluation, but it has been suggested that there may be a reciprocal connection as well, such that low self-appraisal (in the absence of illness) makes the subsequent onset of depression more likely. A prospective study, using a community sample of 376 women, provided data about clinical state over a period of 18 months, and self-appraisal questionnaire scores were determined on two occasions separated by 6 months. There was no evidence that low self-evaluation predicted future episodes of depressive illness, except in women who reported previous psychological episodes for which they had sought medical help, and, even for those with previous episodes, much of the predictive power of low self-esteem was accounted for by individuals who were subsequently recognised to have been in the early stages of illness. Conversely, there was little evidence that prior episodes predicted future illness in people with high self-esteem. One explanation of the findings is that recurrent episodes of illness cause progressive impairment of self-appraisal, but other possibilities are also considered. Women who had recovered from illnesses detected at the first interview still had significantly less self-confidence 6 months later than those who were well throughout.