To solve the problems of numerous influencing factors, such as the high uncertainty and leakage risk of gas production pipelines in high-sulfur gas fields, a dynamic analysis of a gas production pipeline’s leakage risk using a dynamic Bayesian network is proposed. By means of Bow-tie model analysis, the primary risk sources of gas pipeline leakage and different accidents are summarized. A temporal dimension was introduced to construct a dynamic Bayesian network model, utilizing the Leaky noisy-OR gate model to rectify and compute conditional probability, thereby facilitating dynamic risk prediction of gas pipeline leakage. Taking the first section of the pipeline of a municipal gas collection station as an example, with the help of GeNIe 4.0 Academic software, the influence degree of each basic event on pipeline gas leakage was revealed. The change curve of gas leakage probability over time was drawn, and the occurrence probability of potential consequences of accidents was computed. The results indicate that the status of flanges, valves, and pipelines are key factors in determining the occurrence of gas leakage accidents, and six risk sources, including medium corrosion in gas leakage accidents, were determined, with these having practical conspicuousness for strengthening the leakage protection of gas pipelines and providing proper support for the formulation of relevant safety measures.