Stress corrosion cracking of pipelines assume two different forms: high pH and near-neutral pH SCC. Many environmental material and loading factors affect the occurrence of either form of SCC at a given location. A complete model encompassing all the factors has been difficult to achieve because of the interactive nature of many variables. A Bayesian network-based model is useful in this context because of its flexibility and its ability to learn from observations. The results of model using pipeline data will be presented. The strengths of the approach and the gaps in knowledge will be discussed.