The pullout resistance of a soil nail is a key parameter in the soil nailing design. Many factors have influences on the pullout resistance, such as overburden stress, grouting pressure, soil dilation, and degree of saturation of soil. These factors are not well considered in the current design methods. In this study, Bayesian probabilistic approach is employed to build a new predictive formula for estimation of pullout resistance of grouted soil nails. Firstly, the Bayesian model class selection framework is briefly reviewed, and specific expressions are derived for the multivariate linear regression formula. Then, a series of laboratory soil nail pullout test data under different combinations of overburden pressure, degree of saturation and grouting pressure are analyzed by Bayesian method. The analysis shows that grouting pressure and degree of saturation are the most important factors in the estimation of soil nail pullout resistance. It is found that the proposed model is better correlated with experiment results than Hong Kong design method.