Quantitative risk assessment due to external corrosion is a challenging task for pipeline engineers because of the uncertainty in data related to environmental and physical variables such as soil type drainage soil chemistry CP effectiveness coating type and coating properties. In addition most methods or standards employed in practice to compute a corrosion rate assume a constant rate that does not change with the variables though the variables change in time and in location along the pipeline and thus cannot be used to assess risk variability in different sections of a pipeline.In this presentation a risk assessment methodology based on Bayesian network models and its applicability is demonstrated. The model is developed by combining an exhaustive database of expert knowledge empirical field data and mechanistic knowledge of the process. A case study done on an oil pipeline in eastern China is discussed. The predictions of the assessment model are validated with multiyear ILI data. The model and the ILI data are in good agreement. The validated model was used to identify the most probable factors in external corrosion of this pipeline. In this case the most probable factor in corrosion was poor original construction quality of the pipeline which led to dents and degradation of the poor quality asphalt coating on the pipeline. The approach to combining the model with the ECDA is also discussed.