The subcortical dysfunction hypothesis of verbal learning and memory deficits in depression was evaluated by comparing the memory test profiles of unipolar depressives (n = 40) and bipolar depressives (n = 9) with those of patients with a prototypical subcortical dementia (Huntington's disease, HD), patients with a prototypical cortical dementia (Alzheimer's disease, AD), and normal controls. In a discriminant function analysis that well-differentiated the HD, AD, and normal subjects, it was found that 28.6% of the depressed patients were classified as HD patients (DEP-HD subjects), 49.0% were classified as normals (DEP-N subjects), none were classified as AD patients, and 22.4% were not well-classified. The DEP-HD group closely resembled the HD group on additional indices of verbal learning and memory, and differed from the DEP-N group, which strongly resembled the normal control group. DEP-N patients also performed significantly better than DEP-HD patients on a number of other neuropsychological tests (e.g., WAIS-R Digit Symbol, category fluency, Trail Making Test Part B). The findings provide support for the subcortical dysfunction hypothesis, but only for a subgroup of depressed patients. Implications for differentiating depressive "pseudodementia" from AD are discussed.