O'Brien's two-sample tests are the extensions of the conventional tests for assessing the differences between treatment groups which take account of the possible heterogeneous nature of response to treatment and which might be of use in the reorganization of the subgroups of patients for whom the experimental therapy might have most advantage.