Depression and anxiety have been linked to elevated inflammation in cross-sectional and longitudinal studies. Yet, in terms of longitudinal studies, findings are inconsistent regarding whether depression predicts worsening inflammation or vice versa, and anxiety has been infrequently examined. Further, we know little about longitudinal relationships between inflammation and specific symptom profiles of depression and anxiety. The current study examined longitudinal associations between depression and anxiety symptoms and inflammation in 13,775 people (59% women, average age = 67) participating in the Health and Retirement Study - a population-based study focused on older adults. High sensitivity C-reactive protein and depression and anxiety symptoms were measured at two time-points separated by four years. We used cross-lagged panel models to examine bidirectional relationships, and tested interactions with gender. We found that depressive symptoms predicted increasing inflammation for men, but not for women, and inflammation predicted worsening depression for women, but not for men. These gender differences were driven by somatic symptoms. Specifically, somatic symptoms predicted increasing inflammation for men only and were predicted by inflammation for women only. Regardless of gender, inflammation predicted worsening dysphoric symptoms of depression, and lack of positive affect predicted increasing inflammation over time. Anxiety was not associated with inflammation longitudinally. These findings indicate bidirectional relationships between depressive symptoms and inflammation, but not between anxiety symptoms and inflammation, and that the direction of these effects may differ by gender and type of depressive symptom.