Abstract: Reasonable and accurate forecast of city gas load has become an important guarantee for government policy formulation, gas source terminal and pipeline network construction, as well as the optimized operation and scheduling of natural gas. Based on extensive research on relevant literature and practice at home and abroad, the research progress of city gas load forecasting technology was systematically reviewed in the light of the development history of the natural gas industry. The forecasting methods were divided into the early traditional forecasting methods, machine learning-based forecasting methods and combined forecasting methods. The research on applying intelligent algorithms based on machine learning to forecast natural gas loads has achieved the leapfrog development in China and thus the research group becomes the most active internationally at the current stage. The combined forecasting methods have refined the analysis on natural gas consumption characteristics and allow for more accurate forecasting results. Hence, the research on the combined application of forecasting methods has become a hot topic in current gas load forecasting research. In the future, more attention should be paid to the application of big data processing and advanced algorithm technology in gas load forecasting, but the in-depth exploration of the technical essence, such as the load formation mechanism and characteristics, should not be neglected.