In the estimation of the ionospheric total electron content from the Global Positioning System (GPS) observables, various instrumental systematic effects such as the biases in the GPS satellites and receivers must be modeled. This paper describes a procedure, based on a Kalman filtering approach, for estimating these instrumental biases as well as the total electron content at each GPS station, using dual GPS data. The method is applied to six data sets of 48 hours each, spanning one year from the Deep Space Network with GPS stations in Australia, Spain, and the United States. The formal errors for the estimated satellite biases and for the total electron content at each station are about 0.07 ns and 0.2¿1016 el/m2, respectively. The variation in time of the satellite biases (relative to the mean of all of them) estimated in different epochs during 1-year period, is below 1 ns.