Background: Generalized anxiety disorder (GAD) shows the weakest treatment response among anxiety disorders. This study aimed at examining whether an acceptance-based group behavioral therapy (ABBT) for patients in a Brazilian anxiety disorders program, combining mindfulness and exposure strategies, can improve clinical outcome when compared with a standard nondirective supportive group therapy (NDST). Methods: Ninety-two individuals diagnosed with GAD were randomized to receive 10 sessions of either ABBT or NDST. Assessments at pretreatment, midtreatment, posttreatment, and 3-month follow-up comprised the following outcome measures: Hamilton Anxiety Rating Scale (HAM-A), Penn State Worry Questionnaire (PSWQ), Depression Anxiety and Stress Scale (DASS), and the Clinical Global Impressions (CGI). The World Health Organization Quality of Life (WHOQOL) was administered at pretreatment and posttreatment. Results: The mixed-effects regression models for DASS-stress, Hamilton Anxiety Interview, and CGI showed a significant effect for Time and the Time × Treatment effect, but not for the Treatment main effect. Similarly, there was a significant Time × Treatment effect for the PSWQ, but not main effects of Time or Treatment. Altogether, these data indicate that symptoms decreased in both conditions across treatment and follow-up, and that the rate of change was more rapid for those participants in the ABBT condition. We found no differences between groups from pretreatment to posttreatment in DASS-anxiety or any secondary outcome measure, but for the physical health domain of WHOQOL, which was faster in ABBT. Conclusions: Both groups showed good clinical outcomes, but in general, participants of the ABBT group improved faster than those in the NDST group.