On-shore oil and gas pipelines have been known to exhibit stress corrosion cracking (SCC) under certain combination of circumstances. Two forms of SCC are recognized - high pH SCC that is intergranular in nature and near-neutral pH SCC that is transgranular in nature. Several factors contribute to SCC, including coating type, coating disbondment, applied cathodic potential, soil type, pressure fluctuation, and steel surface condition under the coating. Typically, SCC prediction involves considering one or two of these factors at the exclusion of others. In this paper, we consider all the relevant factors in a Bayesian network framework. The existing knowledge of cracking mechanisms are incorporated as conditional probability tables. The model predictions are compared to industry experience. The effect of improving knowledge and data acquisition on predicted probability are evaluated.