A procedure, based on the quantile function, is developed for testing the null hypothesis for continuous univariate distributions depending upon a finite parameter vector , for type II censored data sets. The statistic is obtained from a regression point of view by applying discrete time regression techniques to the sample quantile function. The resulting test statistic is computationally quite simple. Further, under both the null and local sequences of alternatives, it is asymptotically equivalent in distribution to the likelihood ratio statistic. A simulation study indicates that the statistic converges rapidly in distribution to a chi-squared random variable.