In this paper we contribute to the debate on the practice of presidential veto and we aim to empirically investigate what are the determinants of its occurrence. We focus on veto practice developed under provisions of current Polish constitution. Using web scraping techniques we collected data on 5050 bills passed by the Parliament or vetoed by the president in period 1998-2022 and merged them with presidential and governmental characteristics. Our empirical approach consists of two econometric models, i.e. a probit regression model estimated using the generalised linear model function and decision tree model. The obtained results indicate several significant determinants of presidential veto occurrence, i.e. the person of bill's initiator, cohabitation, president's term in office, and the age of president. The outcomes of our analysis entail numerous policy implications related to inter alia assessment of functioning of presidential veto in constitutional system, and its impact on the quality of state's governance.