Cognitive-behavioral therapy (CBT) is an evidence-based treatment for anxiety; however, a growing body of research suggests that CBT effect sizes are smaller in Veteran samples. The aim of this study was to perform secondary data analyses of a randomized controlled trial of CBT for late-life generalized anxiety disorder compared with treatment as usual (TAU) in a Veteran (n = 101) and community-based (n = 122) sample. Veterans had lower income and less education than community participants, greater severity on baseline measures of anxiety and depression, poorer physical health, and higher rates of psychiatric comorbidity. Treatment effects were statistically significant in the community sample (all ps < 0.01), but not in Veterans (all ps > 0.05). Further analyses in Veterans revealed that poorer perceived social support significantly predicted poorer outcomes (all ps < 0.05). Our results underscore the complexity of treating Veterans with anxiety, and suggest that additional work is needed to improve the efficacy of CBT for Veterans, with particular attention to social support.