Abstract
Test procedures in clinical dose finding studies are considered when it can be assumed that increasing dose levels lead to a larger or at least equal efficacy. If one is interested in various comparisons within the set of dose levels, one arrives at a problem of multiple testing against ordered alternatives. Some multiple test procedures controlling the multiple level of significance are compared in a simulation study. It follows that a simple procedure that only uses comparisons between neighboring dose levels can be recommended. The degree of approximation of this Bonferroni-type procedure is investigated under normal assumptions.