Abstract: A class of nonparametric tests for comparing two survival distributions and based on randomly right censored data is proposed. The tests developed assume neither the proportionality of the hazard functions of and nor the equality of the censoring distributions. A subclass of the proposed tests compares favorably to the Gehan and logrank tests for testing standard hypotheses. It can be used to test hypotheses concerning F and G over a prespecified interval.
Key words and phrases: Censored data, crossed survival distributions, quantile.