Most patients with diagnosable major depression may fail to show the classic symptom pattern for which standard assessment instruments were designed. To investigate the influence of subtype of depressive illness on Zung Self-Rating Scale (SDS) and Hamilton Rating Scale (HRS) validity, 36 in-patients, who met Diagnostic and Statistical Manual (DSM-III) criteria for major depression (47% male, mean age of 42 years), were assessed on the HRS and the SDS on the same day at baseline, and at the end of each successive week of treatment, to total 164 paired assessments. The HRS was rated by two pscyhiatrists, while the SDS was factor analyzed. An analysis of the results showed that across all paired assessments, the correlation between the two scales was statistically significant and of moderate strength. Correlation between HRS and SDS scores declined sharply with increasing severity of depression as measured by the HRS, so that there was no positive relationship between clinicians' and patients' ratings of their depressive severity. The factor analysis of SDS indicated a stable, underlying factor structure in the symptom patterns of a wide range of depressive disorders. Agreement between the two scales showed that the endogenicity factor was most significant in predicting agreement, with lower agreement for nonendogenous symptoms. To the extent that conventional depression rating scales are unsuited to those not in the melancholic subtype, alternative assessment measures, reflecting changing boundaries of diagnosis, need to be applied. (BL)