AbstractThe present research investigated the extent to which mild depressives and normals differed in their self‐referent processing of personal information. In Experiment 1, these subjects made two types of ratings on depressed (e.g., bleak, dismal) and nondepressed (e.g., loyal, organized) content personal adjectives. Half of the adjectives in each content category were rated for a semantic attribute (Does this word have a specific meaning or relate to a specific situation?), whereas half were rated for degree of self‐reference (Does this word describe you?). These ratings were followed immediately by an incidental recall task, in which subjects recalled as many of the adjectives as possible. Consistent with predictions generated from a content‐specific self‐schema model, normals displayed superior recall for self‐referenced nondepressed content adjectives, when compared to recall for self‐referenced depressed content adjectives and recall for semantic ratings (both depressed and nondepressed content). In contrast, mild depressives exhibited enhanced self‐referent recall for both types of content, when compared to their recall for semantic adjectives. This finding suggested that mild depressives utilize a self‐schema which incorporates both depressed and nondepressed content. Experiment 2 explored this suggestion further by substituting an other‐referent rating task (Does this word describe Pierre Trudeau?) for the semantic judgment used in Experiment 1. Again, consistent with a content‐specific self‐schema model, normals displayed superior recall only for self‐referenced nondepressed adjectives. Mild depressives, however, showed enhanced self‐referent recall, relative to other‐referent recall, only for depressed content adjectives. For nondepressed content, mild depressives did not distinguish between the self‐ and other‐referent conditions. This finding hinted that the nondepressed component of the mild depressives self‐schema may operate at a somewhat reduced effectiveness, but only when required to diff