Leaks and breakdowns of pipelines can lead to catastrophic failures and cause economical losses worldwide. Currently, condition monitoring has become a challenging process because of various reasons such as fluctuating external conditions, natural hazards. Pipelines are installed in severe conditions and are subjected to degradation mainly due to corrosion and metal loss. This study attempted to classify different types of metal loss faults using historical inspection data of oil and gas fields. For this purpose, Support Vector Machines (SVM) were employed to classify and predict various types of metal loss faults which were affecting the life condition of a crude oil pipeline. The historical inspection data was acquired from a crude oil pipeline located in Sudan. Different types of SVM models were trained and quadratic SVM type was selected for the present study due to its high prediction accuracy. The performance evaluation of the proposed SVM model was done using the confusion matrix. The developed SVM model provides promising results with a prediction accuracy of 93.0%. As a result, the fault detection rate (FDR) for all faults is found to be 90.4%, while the misclassification rate (MR) is 9.6%. The prediction of metal loss fault type may help in condition assessment and maintenance schedule to take prior actions for the better life of pipeline which reduces the degradation rate of a pipeline.