As the foundation of material and energy flows in wetland ecosystems, the net primary productivity (NPP) is one of the most important research hotpots in the terrestrial ecosystems carbon cycle. The estimation methods of vegetation NPP in the wetlands have been developed from the sampling observation to model simulation. The models could be usually classified into statistical models, light utilization efficiency models and process models. With the development of computers and remote sensing technologies, remote sensing data have been widely applied in estimating regional vegetation productivity. As shown in some existing review papers, however, the traditional classification could not completely contain all the estimation models of vegetation NPP, especially for the simulation models using remote sensing technique. This paper reviewed the NPP estimation models of almost all kinds of the vegetation in the wetlands, and especially introduced some most popular remote sensing models. The statistical models were classified into climate-productivity relationship models and remote-sensing-productivity relationship models, and the process models consisted of single process models, multiple process coupled models and remote sensing-process coupled models. In the paper, the research advances in the models for estimating NPP in the wetlands were summarized, and the limitations of these models were also discussed comprehensively and systematically. The future perspective on models research could focus on improving the accuracy of the field measurements, reducing the uncertainties of simulation, and using the multi-sources data. It is remarkable that coupling remote sensing data with process models by the data assimilation methods would become one of most important development aspects for NPP estimation of the wetlandvegetation.