The capacity to deliver some forms of behavioral treatment via computers may prove to be a small revolution in the delivery of mental health care. Although early research on the efficacy of these approaches has yielded mixed results, this new strategy offers tremendous potential to provide empirically supported therapies to many individuals who would never access psychiatric care, to extend the time and expertise of clinicians, and to offer improved care and monitoring. However, the great promise of computer-assisted therapies may be diminished if their benefits are overstated or if they are broadly released or disseminated before being carefully evaluated using the same methodologic standards that are requirements for evaluating clinician-delivered therapies. In this article, we review the current status of empiric support for computer-assisted therapies, advocating for enhanced rigor to identify those that are most effective, as well as the need to more thoroughly assess possible adverse effects, recognizing that even a modestly effective computer-assisted intervention could have enormous impact.