In this paper, a model is established and weight functions are given for robust batch processor in satellite precision orbit determination. The effects no robustuess weight functions. It is shown that the robust batch processor can give the best estimate results from normal observations by making full use of effective observations, finding limitations of useable observations and eliminating deleterious gross errors. The robust computations steps are also given. Finally, satellite laser ranging (SLR) data of satellite Lageos were processed, the satellite orbit estimated from the robust estimation is more stable than that from least square estimation and this verifies the good robustness of robust batch processor in practice.