This study presents an approach based on the Gauss-Markov Kalman filter to assimilate the total electron content observed from ground-based GPS receivers and space-based radio occultation instrumentations (such as FORMOSAT-3/COSMIC (F3/C) and FORMOSAT-7/COSMIC-2 (F7/C2)) into the International Reference Ionosphere. Observing System Simulation Experiments (OSSEs) show that the data assimilation procedure consisting of the forecast and the measurement update steps can better improve the accuracy of the data assimilation analysis than the assimilation procedure using the measurement update alone. Compared with F3/C, the denser F7/C2 occultation observations can improve the analysis accuracy significantly as suggested by OSSEs. The real data assimilation results are further validated with global ionosphere maps, the global ground-based GPS measurements, and the ionospheric F2 peak height and electron density sounded by ionosondes. Both the OSSEs and validation results confirm that a number of improvements to the data assimilation procedure presented in this paper can indeed be used to reconstruct the three-dimensional ionospheric electron density adequately.