Abstract: In Exp I, 60 undergraduates' mood responses to manipulated expressions were assessed. Ss then read 2 passages of the same emotional content, either anger-provoking editorials or humorous selections by Woody Allen. Later, one selection was recalled while frowning and the other while smiling. In the self-produced cue group (i.e., those whose mood was affected in the original assessment), recall was significantly better for the editorials when frowning and for the Woody Allen stories when smiling. In the nonself cue group, expressions did not affect recall. In Exp II, 20 undergraduates heard and recalled 3 kinds of sentences—angry, sad, or fearful—in the same 3 expressions. Again, only in the self-produced cue group was recall significantly better when sentence and expression were consistent. Findings indicate that these effects are due to mood rather than expression and to the match between expression and content at the time of recall.