In this study, we applied a stochastic sea ice generator and automatic ice routing to investigate navigation conditions in the Arctic. The stochastic generator provides long (over 100 years) interconnected spatial-temporal fields of sea ice parameters that were used as a source of information to calculate navigation routes using the optimal ice routing tool. A stochastic sea ice generator is a probabilistic model that reproduces synthetic sea ice data. Temporal connectivity of generated data is provided by Markov chains, while spatial connectivity is achieved using empirical random probability fields. To determine the matrices of transition probabilities for Markov chains and empirical random probability fields of sea ice concentration, we used satellite data from the OSI SAF project for the last 20 years. The grid area of the probabilistic model covers the Barents Sea, Kara Sea, and nearby waters with a spatial resolution of 10 km. Time step of the model is one day. We tested the stochastic generator by comparing its results with statistical characteristics of ice concentration obtained from satellite data of the OSI SAF project. The mean absolute error of the statistical characteristics of ice concentration was found to be less than 5%. As an example of the practical use of the stochastic generator, we used ice routing tool to calculate the probability of navigation start at the line from Murmansk to Pobeda field (Kara Sea) on different calendar dates under the assumption that the vessel can operate in open water or very open ice only. The average and extremely early and late dates were estimated.