As inexpensive interventions gain empirical support, there is an increasing risk that such data may be used by health maintenance organizations to unfairly restrict the number and type of therapy sessions reimbursed for all clients, even those less likely to benefit from economical treatments. As a result, it is important to identify clients who may not respond to specific therapies and to empirically support ways to treat them. Successful treatment of nonresponders is also valuable because predictors of treatment failure tend to predict cost related to medical and disability expenses. Using generalized anxiety disorder as an example, this article suggests a flexible and comprehensive approach to cost-benefit analysis in psychotherapy that includes clients who may not improve in response to current data-based interventions. In addition, suggestions are made for the identification of alternative treatment approaches, and a potential treatment allocation model is recommended.