As a general term that describes the accumulation of organic matters within specific temporal and spatial scopes on crop land, woodland, grassland, or other types of lands, land net primary productivity (NPP) is considered an important parameter to measure carbon cycle, guide land use, assess ecological security, reflect environmental changes, and indicate the level of food security. The estimation precision of NPP is significantly influenced by the type of models and input of key surface parameters of ecosystems. In recent years, with the continuous growth of remote sensing data and the rapid development of remote sensing data processing technologies, NPP estimation models based on remote sensing data, as compared to NPP estimation using traditional observation data such as climate and soil data with coarse spatiotemporal resolutions, have become very prominent in analyzing temporal and spatial heterogeneity. Based on the Web of Science and CNKI databases and statistical analysis methods, this study systematically reviewed research on NPP and its estimation models integrating remote sensing data in China and internationally. The commonly used models can be divided into four categories: statistical models, light use efficiency models, process models, and coupling models. We examined the mechanisms, differences, suitability, and limitation of the various kinds of models, Based on an analysis of the difficulties and scientific challenges that face integrating remote sensing data into NPP estimation models, research prospects are put forward with regard to model mechanism, influencing factors, data provision, parameter derivation, expansion of spatiotemporal scales, and hardware and software supports.