Abstract: Objective: A generalized non-central procedure is proposed for covering censorship. It is derived from its classical counterpart and appllied in determining the sample size required in the poly-sample survival rate test. Methods: The first step inherites the classical procedure for determining the required homogenetic effective sample size based on existing tables of the noncentral chi square distribution and expression for the non-centrality parameter of the classical r×2 chi square statistic. The second step yields the required sample size at a given censoring rate, which is operated inversely from the parametric expression of the homogenetic effective sample size under Weibull survival distributions. The sample allocations are realized by the item-by-item iteration. Results: By contrast with the existing procedures for sample size determination adjusted censoring, the procedure has such characteristics as that it is matched with the poly-sample survival rate test, that it is free from the assumption of exponential distributions, that it reduces to its classical counterpart when there is no censoring, and that the observed power of the test coincides with the prescribed power. Conclusion: The procedure can be applied to planning poly-sample clinical cancer studies. A worked example illustrates the planning process.