The up-and-down method in bioassay using multiple samples in each trial is investigated. The bias and precision of the estimators for ED50 applying the dose-averaging formula and probit analysis are computed and relatively efficient sampling schemes are discussed. The use of the method to estimate extreme percentage doses is developed. It is shown that the estimators from this method are generally more efficient than the non-sequential estimators and are as good as other comparable sequential estimators. Several numerical examples of the asymptotic distribution of the dose-levels from this sampling procedure are tabulated.