Abstract: With reference to the accuracy and time of rapid and ultra-rapid satellite orbit and the unbalanced distribution of global tracking stations, GNSS Data Analysis Centers are meeting with big challenge. This paper proposes an optimal stations selected model called SSS (selected step by step) which is based on the GDOP (geometric dilution of precision) value of observation equation. Firstly, the calculation methods of optimal and the minimum of ground tracking stations for precise orbit determination were deduced. Secondly, according to the global grid of s°×s°and k°×k°, the distribution of minimum stations were selected out. Thirdly, based on the standard of minimum GDOP, an optimal distribution of global tracking stations was gradually accumulated step by step. Six days continuous experiment shows, on the same numerical computation ability, SSS model can reach 90% level of accuracy as the whole data processing and reduce computation time at less 50%. Comparing with the ordinary methods, it shows that SSS model can get the same accuracy as the ordinary methods, while could save time reaching up 20%. Moreover, several comparison experiments indicated that SSS is the optimal or sub-optimal model for the station selected and improves the efficient of data analysis centers.