Theories indicate that individuals with generalized anxiety disorder (GAD) avoid imagining threatening scenarios that have not occurred. Cognitive exposure to these images is a component of treatment; however, few studies have examined its efficacy. The current study assessed the impact of cognitive exposure and varying exposure content on GAD symptoms and cognitive processes. Forty-eight individuals with GAD were assigned to three sessions of : (1) variable exposure (VE), (2) consistent exposure (CE), or (3) neutral control writing (NC). Emotional activation was assessed during each session. Outcome measures were administered at pretest and 1-week follow-up. The CE condition showed improvements in GAD symptoms and cognitive processes, the VE condition showed less belief that their worse worry would occur, and the NC condition showed reduced GAD symptoms and interpretation bias. Emotional activation decreased across session in the CE condition; however, this did not predict outcomes. The theoretical and clinical implications are discussed.