In a survey of 200 depressed patients, clinical symptoms of depression were assessed by both semistructured clinical interview and self-report. Both instruments were given during the acute episode and at follow-up, ten months later, when a majority of patients had recovered. Results indicate concordance between clinical assessment and self-report is low during the acute episode but generally improved at follow-up. Ratings of individual symptoms show a fair degree of specificity. The overall assessment of severity during the acute episode is particularly discordant. These findings demonstrate that self-reports are useful in measuring the presence or absence of symptoms and therefore valuable in assessing recovery. However, self-report ratings from acutely depressed patients are not a reliable estimate of the severity of their symptoms.