There are known to be enormous Cenozoic coal-type oil and gas resources located in the basins of the South China Sea, among which the Paleogene coal-measure source rock are one of the main source rock. In order to more effectively analyze the distribution laws of coal-measure source rock in marginal sea basins and guide coal-type oil and gas explorations, the Oligocene coal-measure source rock in the Qiongdongnan Basin were selected as examples in order to systematically analyze the types, development characteristics, control factors, and distribution prediction methods of coal-measure source rock in marginal sea basins. The Qiongdongnan Basin is located in the northern region of the South China Sea. Previous explorations of the area have determined that the Oligocene coal-measure source rock in the Qiongdongnan Basin have typical “binary structures”, which include coal seams and terrigenous marine mudstone. Among those, the terrigenous marine mudstone has been found to greatly expand the scope of the coal-measure source rock. In addition, the coal seams which have been exposed by drilling have been observed to have the characteristics of thin single layer thickness, many layers, and poor stability. Meanwhile, the terrigenous marine mudstone has the characteristics of large thickness and wide distribution. The development of coal-measure source rock is known to be controlled by many factors, such as paleoclimate and paleobotany condition, paleo-structure and topography, paleo-geography, rise and fall of base level, and so on. In accordance with the comprehensive analyses of various control factors of coal-measure source rock, and the changes in water body energy in sedimentary environments, a genetic model of the “energy belt” for the development and distribution of coal-measure source rock was proposed. Also, the development and distribution characteristics of coal-measure source rock in different types of energy belts were clarified. Then, based on the development and distribution characteristics of coal-measure source rock and their controlling factors, prediction methods of the distribution ranges of coal-measure source rock were proposed from both qualitative and (semi-) quantitative aspects. That is to say, a method for (semi-) quantitative predictions of the distribution ranges of coal-bearing intervals based on model wave impedance inversion and neural network wave impedance inversion, along with a method for (semi-) quantitative predictions of the distribution ranges of terrigenous marine mudstone based on topographical slopes, were introduced in this study.