We consider Bayesian estimation of a sample selection model and propose a highly efficient Gibbs sampler using the additional scale transformation step to speed up the convergence to the posterior distribution. Numerical examples are given to show the efficiency of our proposed sampler.