Purpose: While new drug therapies offer one or more health benefits to patients, they also frequently pose a risk of adverse events. Therefore, an explicit, quantitative method for undertaking harm-benefit analysis is required to aid in therapeutic decision making. The main objectives of this study were to 1) identify criteria for an ideal harm-benefit evaluation method, 2) assess all currently proposed harm-benefit analysis methods for therapeutic interventions against these criteria, and 3) from these results, recommend the optimal method. Methods: Specific criteria considered essential for a valid, practical, quantitative harm-benefit analysis method were developed from documents published by the US Food and Drug Administration, the Council for International Organizations of Medical Sciences, and other published literature. Next, we identified harm-benefit analysis methods proposed in the published literature by searching all major medical journal electronic databases including Medline, EMBASE, EBM Databases, PapersFirst, and ProceedingsFirst, and through a secondary scan of the reference lists of papers identified in the primary search. The analytic methods were then evaluated against the proposed criteria and a recommendation for the optimal methodologic framework was made. Results: We developed ten specific criteria which we propose are essential for a valid, quantitative harm-benefit analysis. Twelve harm-benefit analytic methods were identified from 224 published articles plus additional textbooks and working papers and were evaluated against the ten proposed criteria. Five of the methods met less than six criteria, six methods met between six and eight of the criteria, and only one method, the incremental net benefit framework, met all ten criteria. While three methods permit the simultaneous consideration of multiple harms and multiple benefits, the incremental net benefit framework also facilitates the incorporation of the uncertainty around the incremental harms and benefits, and the inclusion of decision-maker's risk preferences. Conclusions: The majority of proposed harm-benefit analysis methods are limited in that they do not facilitate the simultaneous evaluation of multiple harms and benefits or incorporate uncertainty into the analysis. The incremental net benefit framework met all ten proposed criteria and should be further developed as a quantitative tool for the analysis of the potential harms and benefits of new therapeutic interventions.