A group randomized trial of adding a home-based walking program to a standard group cognitive behavioral therapy (GCBT+EX) was compared with groups receiving GCBT and educational sessions (GCBT+ED). The study was implemented in an outpatient clinic providing GCBT for clients diagnosed with panic disorder, generalized anxiety disorder or social phobia. Pre- and post-treatment measures included the self-report depression, anxiety, and stress scale (DASS-21) and measures of physical activity. From January 2004 to May 2005, six groups were allocated to GCBT+EX (n=38) and five to GCBT+ED (n=36). Analysis of covariance for completed cases (GCBT+EX, n=21; GCBT+ED, n=20), adjusting for the group design, baseline DASS-21 scores, and anxiety diagnosis showed significant effect for GCBT+EX on depression, anxiety, and stress (regression coefficients=-6.21, -3.41, and -5.14, respectively, p<0.05) compared to the GCBT+ED. The potential of exercise interventions as adjunct to GCBT for anxiety disorder needs to be further explored.