The transport network in the regions of the North of theRussian Federationbasically remains seasonal (waterways, winter roads). The duration of river shipping season, depending on the climatic conditions, is 110–160 days, and the time of operation of winter roads varies within 120–210 days. Under these conditions, the accuracy of predicting the beginning and end of shipping season for the northern rivers plays a very important role. The article proposes a method for forecasting the duration of ice phenomena in the areas of shipping routes based on the use of the mathematical apparatus of Markov chains. An estimate of probability of an accurate forecast is given, taking into account the conformity with Bayes theorem and related dependencies. Verification of the method on the basis of real data proved that the forecast accuracy and probability of its implementation were sufficient for timely and effective organisation of preparatory operations for next shipping season on northern navigable rivers.