Production profile of super-tight or shale gas wells is mainly influenced by the reformed fractures and leads to a typical characteristic decline in production is rapid at the beginning of the decline period, after which it slows down until the production levels off. Reservoir engineers focus on how to predict accurate future production and estimated ultimate recovery (EUR) in the presence of a short production history of shale gas well through simple and convenient means. Numerical and analytical methods are not always accurate enough owing to accurate fracture and reservoir parameters, which are complex and challenging to be obtained efficiently. Therefore, empirical methods have been widely introduced in the industry owing to their simplicity and efficiency. While traditional Arps's decline analyses are not suitable for super-tight or shale gas wells as these models are based on convention reservoirs that are mainly dominated by boundary dominated flow, then several empirical methods targeting tight or shale gas wells have been proposed. The most widely used empirical methods are Valko's stretch exponential production decline (SEPD) and Duong's rate decline for fractured dominated reservoir (Duong's method) but with some limitations. This study analyzes the basic theories of Arps's methods, SEPD, and Duong's method in detail to ascertain the reason behind their limitations. Subsequently, a new empirical method is proposed to estimate reliable future production and EURs for fracture-dominated reservoirs. The new methodology is based on the empirical relationship between production and time in a fracture-dominated flow regime, and considers the influence of time on the fracture time exponent. The proposed method is compared to SEPD and Duong's method through theoretical analyses and different empirical applications of shale gas wells in the Sichuan Basin (China). The results have further proved that SEPD underestimates the outcomes, particularly for shale gas wells with low productivity, and Duong's method overestimates the outcomes and is significantly influenced by an irreversible decline in the slow decline stage. Meanwhile, the comparisons have shown that the proposed approach performs better than the other two methods in both production forecast and EUR estimation, in both short and long production histories.