In this paper we present the usage of neural networks and hidden markov models for learning driving patterns. We used neural networks for short-term prediction of lateral and longitudinal vehicle acceleration. For long- time prediction, hidden markov models provide recognition of individual driving events. The experiments performed showed that both techniques are very reliable. Recognition rate for driving events is above 98% and prediction error for events in the near future is very low. Predicted events will be used to support drivers in solving guidance navigation tasks.