We apologize for not having used the phrase "random assignment" when describing our double-blind treatment methods. Indeed, the patients were assigned to the three treatments according to a random-assignment schedule, which resulted in three treatment groups of equal sizes. In terms of potential predictors of a successful taper, Dr. Gupta et al. wonder why we did not use the level of baseline symptoms as a potential predictor. We did not do so because depressive symptoms at baseline did not correlate significantly (p<0.10) with treatment outcome and, thus, as we stated in the article, did not fulfill our criteria for variables entered into the logistic regression analysis. We believe that the other comments made by Dr. Gutpa and colleagues, as they relate to the study group, the generalizability of the results, benzodiazepine dependence, long-term treatment for generalized anxiety disorder, resistance to benzodiazepine therapy, and the effect of treatment group on benzodiazepine status at 12 months, were adequately covered in the article.