Depression therapy literature cites the importance of self-concept in formulating therapeutic strategies. Traditional depression measures catalog a client's symptoms, but fail to tap the structure of dysfunctional thoughts and judgments that form the client's self-concept. Vector-product multidimensional scaling was used to develop a measure of depression defined by the pattern of relationship judgments formed by relating self-concept to other concepts significant in depression. A control group (N=17) and a depressed group (N=10) rated pairs of descriptive self-image concepts from 1 to 9 according to how similar they judged each pair to be. These similarity ratings were analyzed through principal components analysis into factors representing the dimensions underlying each subject's self-concept. Cognitive maps were generated for control subjects, the average control subject, and the average depressed subject. Cognitive maps derived from averaged relationship judgments revealed significant differences between self-concepts of control and depressed subjects; individuals' discriminant function scores correctly catagorized 96% of the subjects into the correct groups. For depressed subjects, the configuration of pair-concept relationship judgments provided insight into the interrelationships between the positive and negative aspects of self-concept which form the unique pattern that constitutes depression. (Author/NRB)