In this paper, we propose the method of statistic analysis in crossover design with bivariate endpoints. The bivariate endpoint contains normal endpoints and binary end-points. In the section of normal endpoints, based on the existence of the carry-over effects, the corresponding method of the hypothesis test and the the sample size calculation axe given. In the section of binary endpoints, based on the Mainland-Gart test, the cor-responding method of the hypothesis test and the sample size or power calculation axe given. In the end, the calculations of sample size determined and power are facilitated by R programs. |