The purpose of this study is to evaluate empirical criticism of the psychometric properties of the Hamilton Depression Rating Scale (HDRS) by undertaking an "internal construct validity" evaluation. Individual symptom responses for the HDRS were obtained from 370 patients meeting Research Diagnostic Criteria for major depression. A multidimensional normal item response theory (IRT) model was used to assess the dimensionality of the HDRS and estimate corresponding item parameters which were then expressed as factor loadings and item thresholds. The analysis yielded a five-dimensional solution. The first dimension (depressed mood, guilt, suicide, work and interests, agitation, psychic anxiety, somatic anxiety, and loss of libido) appeared to define global depression severity. These results are consistent with other published factor analytic studies and provide further justification for thinking that the HDRS total score is a weak index of depressive syndrome severity. The findings provide a benchmark by which the adequacy of future results may be judged, because the multidimensional IRT model does not suffer from the statistical limitations that arise when applying traditional factor analytic methods to discrete symptom ratings.