This paper proposes a data-driven approach in determining an optimal inspection interval for a petroleum pipeline system. The approach accounts for the determination of both the probability of failure and its associated consequences. The probability of failure is estimated by fitting the historical data of failure of the pipeline into either a homogenous Poisson process or nonhomogenous Poisson process (power law). The analysis of historical data reveals the Poissoneous form that gives a better description of the failure process. The consequences of failure are calculated in terms of economic loss, environmental damage, and loss of human life. Both the failure probability and consequences are utilized to estimate the total loss of an operating pipeline system. A risk-based integrity maintenance optimization of the pipeline is achieved by minimizing the economic loss while taking the human risk and maintenance budget as constraints. The proposed framework is utilized in the maintenance planning of a very long cross-country petroleum pipeline system. The outcomes are robust and well validated. The framework can be applied to any engineering system that requires inspection and maintenance planning. © 2013 American Society of Civil Engineers.