This study measures the anticipatory trade creation and trade diversion effects of Customs Union of Belarus, Russia and Kazakhstan (CUBRK) for chosen industries. Those effects are estimated using Gravity Model of Trade (GMT). Log-linerized and exponential specifications of GMT are estimated with the help of OLS and Poisson-Pseudo-Maximum-Likelihood (PPML) estimators. Models include time fixed effects and variables which deal with endogeneity issues. The research data contain large amount of zeros, so OLS is estimated mainly for comparison reasons, since it was proven to be biased in case of zero trade flows. The inference is made based on PPML estimation. In order to capture gradual changes in trade flows, dynamic models are estimated along with static models.