Sediment delivery ratio (SDR) is an indicator to demonstrate the capacity of sediment erosion and delivery of a certain basin. Hence, SDR is usually recognized as a helpful factor when evaluating sediment reduction effects by soil and water conservation management and making policy-decision on ecological construction. Studies on SDR at home and abroad during the last 50 years have been systematically reviewed. Firstly, the essential characteristics of SDR have been stated and discussed. In the expression formula of SDR (S_DR=Y/E), Y is the amount of sedimentation delivery which is equal to the amount of sediment yield or soil loss in a certain watershed. The concepts of Y are different under different circumstances. For example, Y (sediment yield) from the soil erosion perspective is the amount of sediment from the observation cross-section after sedimentation equilibrium at the gully erosion level. But, Y (sediment delivery) from the river engineering perspective is including the erosion-siltation process. Moreover, it is important to note that both definition and quantitative amount between SDR and the rate that sediment from watershed slope returning to channel are totally different, which is because the two mass of sediment are different and could not be used as the same. Secondly, calculation methods regarding to SDR have been summarized and discussed on their advantages and disadvantages, including four types of analogy calculation methods and three forms of modeling methods. Two commonly used soil erosion calculation methods are unit-watershed analogy calculation method and the physical-process modeling method based on the total watershed erosion. Calculation methods discussed in this paper should be used carefully with considering their disadvantages. For instance, disadvantages of the unit-watershed analogy calculation method were observation data only from the unit-watershed, the inadequate representation of larger watersheds and the vogue expression of identifying a unit-watershed with the watershed area less than 1 km². The disadvantage of the physical-process modeling method based on the total watershed erosion was not considering siltation and exchange processes in the slope and channels, such the USLE model. Thirdly, this paper pointed out that the dependence of SDR on scale with its suitable scale domain. On the one hand, the nature of SDR relied on its scale has been discussed and analyzed deeply through the essential characteristics of SDR and scale characteristics of SDR according to many published papers at home and abroad. It was been identified that the scale characteristics embedded in the SDR definition. On the other hand, influential factors of SDR varied and differentiated along with changing of scale domains. These two reasons could explain the calculation method of SDR based on the influential factors for a watershed is hardly to apply on another watershed. Finally, the study on SDR in the future was generally discussed. For instance, it is interesting to revise and set up the calculation methods of SDR to fit different soil and water erosion conditions in China. It also interests to study the dependence of SDR on scale and the scaling methods, which requires long-term temporal and spatial observed data from many different watershed scales around the world. The scaling methods of SDR might be helpful to provide new ways for the research on hydrological process scaling.