We present a fixed sample size procedure and a closed sequential procedure to compare k ( > 1) experimental treatments with a standard. All the treatments produce outcomes with two Bernoulli endpoints. The goal is to select the treatments whose success rates of both endpoints are better than their respective standards. We modify and generalize Dunnett (1984)’s formulation for comparing treatments with one Bernoulli endpoint to a formulation for comparing treatments with two Bernoulli endpoints. We derive the lower bound of the probability of a correct selection for both procedures and use it to calculate the sample sizes needed to achieve probability requirements. An example is given to illustrate the application of our procedures.

線上視訊請點選連結