Previous studies of cognitive specificity in social anxiety and depression have not directly compared data from measures of thought content varying in their relevance to either disorder. The current research compared subjects high in both social anxiety and depression (i.e., mixed) to groups high in only social anxiety or depression, or neither affect, on five cognitive content scales. Negative thoughts specific to depression covaried only with dysphoria, while negative thought content specific to social anxiety covaried with both social anxiety and dysphoria, although both dysphoric and socially anxious groups differed from controls. However, on most cognitive measures the mixed group was the most dysfunctional, replicating Ingram's (1989a, 1989b) evidence of confounding between the affective states of social anxiety and depression. In a second study, the significance of affective confounding in a clinical sample was evaluated by assessing differences in thought content specific to social anxiety among social phobics varying in depressive affect. High dysphoric social phobics reported a lower percent of positive thoughts and significantly more negative thoughts than did nondysphoric social phobics. Limitations of the present research as well as implications for future research on cognitive specificity and affective confounding are discussed.