Twenty-one patients suffering from major depressive disorder are compared for evidence of thought disorder on admission and at the time of discharge. Two aspects of cognition are evaluated: the ability to abstract, and the quality of association. These aspects of cognition were tested using the Shipley-Hartford, the Raven Progressive Matrices, and the Goldstein-Scheerer Object Sorting Test. Althouh the patients showed some improvement in ability to abstract, and although their associations became somewhat less underinclusive or improverished, these improvements did not achieve statistical significance. Consequently, this group of depressed patients did not appear to show any thought disorder of the type under investigation.