The stochastic curtailment for the case of testing a non-inferiority hypothesis in the long-term clinical trials is considered. The way to calculate the conditional powers in the context of censored survival data, normally distributed data, and binomial data will be discussed. The methodology can be easily extended to the location-scale families or asymptotically location-scale families. The projection of stochastic curtailment for non-inferiority case will be superimposed with the superiority case to show the natural relationship between superiority case and the non-inferiority case.