The paper proposes a methodology based on Bayesian Networks for identifying the power two wheeler (PTW) driving patterns that arise at the emergence of a critical incident based on high resolution driving data (100 Hz) from a naturalistic PTW driving experiment. The proposed methodology aims at identifying the prevailing PTW drivers’ actions at the beginning and during critical incidents and associating the critical incidents to specific PTW driving patterns. Results using data from one PTW driver reveal three prevailing driving actions for describing the onset of an incident and an equal number of actions that a PTW driver executes during the course of an incident to avoid a crash. Furthermore, the proposed methodology efficiently relates the observed sets of actions with different types of incidents occurring during overtaking or due to the interactions of the rider with moving or stationary obstacles and the opposing traffic. The observed interrelations define several driving patterns that are characterized by different initial actions, as well as by different likelihood of sequential actions during the incident. The proposed modeling may have significant implications to the efficient and less time consuming analysis of the naturalist data, as well as to the development of custom made PTW driver assistance systems.