This roundtable session will focus on both Bayesian and frequentist approaches to non-inferiority clinical trials. In new clinical trials for anti-infective therapies, there is often historical clinical data available and a potential challenge is to derive a valid method for testing non-inferiority taking the historical information into account. The FDA guidance document regarding non-inferiority compares two standard frequentist methods ("fixed margin" and "synthesis") which both have limitations. To try to solve this dilemma, Viele et al. (2013) have recently presented approaches that incorporate this historical clinical data into an analysis procedure. During the roundtable, dynamic borrowing in the framework of various additional Bayesian hierarchical modeling strategies will be discussed. These additional strategies may include estimating parameters for each trial separately versus pooling, weighting the prior distribution corresponding to each historical study based on the sample size, and incorporating historical borrowing on the control arm by a separate random effect parameter. Furthermore, one critical assumption in non-inferiority trials is constancy; that is, the effect of the control in the historical trial population is similar to the effect in the current active control trial population. The constancy assumption can be at risk due to the potential heterogeneity between trial populations primarily related to subject characteristics, and secondarily to other sources of heterogeneity resulting from differences in patient management (e.g., usage of concomitant medications). As a further refinement, it will be discussed how, in the setting of a non-inferiority trial, a covariate adjustment approach can be implemented to recalibrate the non-inferiority margin based on the effect of active control minus placebo from the current study data. The roundtable session will also discuss additional possible Bayesian approaches in the non-inferiority setting including such topics as power prior models, and hierarchical approaches that could borrow strength from historical controls to reduce the sample size of the active control, etc. in addition to other frequentist approaches such as the TACT (Two-Stage Active Control Testing) method. Finally, the regulatory environment regarding non-inferiority clinical trials (both frequentist and Bayesian approaches) will be examined.