In this study, we explored how moods influence the way people seek information. We conducted a controlled lab study to test our hypotheses drawn from affect-as-information theory. Fifty-eight participants were randomly assigned to the happy or sad condition. They were primed for a certain mood, and they then performed a search task and finished a series of questionnaires. Our findings supported affect-as-information: the comparatively happy participants were inclined to process more general and less specific information; the comparatively sad participants were likely to process more specific information. The findings advances theoretical and empirical understanding concerning the characteristics of users' information seeking behavior under different moods. Our study will contribute to affective search systems design.