The numerous overruns of the investor’s budget during tenders for the construction of railway infrastructure in Poland resulted in the widespread use of a new procedure for awarding public contracts – electronic auction. This procedure has many advantages and potential risks. One of the biggest benefits for an investor is the potential gains from reducing bids. Contractors competing against each other allow for the achievement of optimal prices for the planned construction investment. However, this may cause the originally calculated risks, should they materialize, lead to significant budget overruns. This, in turn, may imply further negative consequences, including exceeding the assumed investment deadlines. The article presents a method of modeling the influence of an electronic auction on a tender procedure with the use of a Bayesian network. Data from completed tender procedures announced by the PKP Polskie Linie Kolejowe S.A. were used to build the network. The created network was then validated, verified and calibrated using new data from 8 tender procedures.