The advancement of pipeline leak detection has been driven by safety, environmental, regulatory, and technological factors across oil, gas, chemical, and broader pipeline industries. This article reviews the historical progression and recent advancement of on-line real-time pipeline leak detection methods from simple pressure drop analysis, to Real-Time Transient Modeling (RTTM), to advanced acoustic leak detection, and to today’s integrated, AI/ML-driven multifunctional systems. While basic pressure monitoring lacks sensitivity and reliability, steady-state simulation models enhanced consistency but failed under transient conditions. Although RTTM incorporating dynamic modeling, computational demands and sensitivity limitations persist. Acoustic leak detection based on negative pressure wave principles offered a fundamentally different approach by directly detecting and precisely locating leak-generated acoustic signals.  Recently, fiber optic sensing technologies have emerged mainly driven by its relative convenience and low cost as an add-on item to an optical communication cable. However, its actual applicability as a standalone leak detection system has proven to be challenging and infeasible owing to the inherited constraints in its reliability in both detecting actual leaks and rejecting false alarm (both false positive and false negative). Each approach built upon the limitations of its predecessors, leading to today’s growing trend toward integrated, redundant, and AI-enhanced systems that offer multifunctional monitoring capabilities beyond leak detection.