Beck's (1967) hypotheses concerning depression build upon notions that a depressed person's cognitions differ from those of nondepressed persons and that a relationship exists in which depressive cognitions assume primacy over mood. A study was conducted to test Beck's hypotheses and to examine the roles that cognition and mood play in depression. The experiment was a 2 x 4 design in which two groups (72 depressed psychiatric inpatients, 61 nondepressed psychiatric inpatients) and 4 conditions were combined factorially. The four conditions were Velten, a cognitive-based condition in which subjects read mood-elating self-referent statements; Comedy Film, a noncognitive condition in which subjects viewed a mood-inducing film; and two control conditions. Pre- and post-tests consisted of the Multiple Affect Adjective Checklist and the Dysfunctional Attitude Scale. The results revealed that dysfunctional attitudes or cognition of clinically depressed subjects differed significantly from those of nondepressed persons and that cognitions became increasingly dysfunctional as level of depression increased. When mood and cognition were both manipulated, cognitions of depressed subjects were modifiable. When mood alone was manipulated, mood changed and subjects became less depressed; no change in these subjects' cognitions was evident. These findings have implications for treatment. (NB)