A comprehensive model to assess crash risks and reduce driver’s exposure to risks on road curves is still unavailable. We aim to create a model that can assist a driver to negotiate road curves safely. The overall model uses situation awareness, ubiquitous data mining and driver behaviour modelling concepts to assess crash risks on road curves. However, only the risk assessment model, which is part of the overall model, is presented in the paper. Crash risks are assessed using the predictions and a risk assessment scale that is created based on driver behaviours on road curves. This paper identifies the contributing factors from which we assess crash risk level. Five risk levels are defined and the contributing factors for each crash risk level are used to determine risk. The contributing factors are identified from a set of insurance crash records using link analysis. The factors will be compared with the actual factors of the driving context in order to determine the risk level.