The factors that maintain generalized anxiety disorder (GAD) symptoms and worry over time are not entirely clear. The Contrast Avoidance Model (CAM) postulates that individuals at risk for pathological worry and GAD symptoms uniquely fear emotional shifts from neutral or positive emotions into negative emotional states, and consequently use worry to maintain negative emotion in order to avoid shifts or blunt the effect of negative contrasts. This model has received support in laboratory experiments, but has not been investigated prospectively in the naturalistic context of daily life. The present study tested the CAM in a longitudinal experience sampling study with a subclinical sample. Participants selected to represent a broad range of symptoms (N = 92) completed baseline measures of GAD and depression symptoms, and eight weekly assessments of worry, experiences of negative emotional contrasts during their worst event of the week, and situation-specific negative emotion. Consistent with the CAM, GAD symptoms prospectively predicted higher endorsement of negative contrast experiences as worst events, independent of depression symptoms. Unsurprisingly, higher negative contrasts predicted higher negative emotion. However, both higher baseline GAD symptoms and weekly worry uniquely moderated (reduced) this relationship, providing consistent support for the idea that worry may blunt the emotional effects of contrasts. Depression symptoms did not have the same moderating effect. These findings support the CAM in an ecologically valid context.