Abstract. To directly describe the relationship between the ERP determination precision and the geometry distribution of GNSS station and satellite, a new dilution of precision factor (DOP) is suggested and named as EDOP. According to the hypothesis of smaller EDOP value corresponding to higher ERP solution precision, the minimum requirement and optimal distribution conditions of station and satellite are obtained to estimate the ERP using GNSS technology. The minimum requirement of ERP solvable are that there are at least two stations and two satellites, and they are not in the same plane. The optimal distribution conditions of station and satellite are that 1) the precision of ERP estimation is highest, when the stations are evenly distributed at the intersection of a sphere and a cone with a cone angle of 109.4712°, and the rotations of the distribution or additions of multiple distributions are still optimal; 2) the higher coverage density of satellites, the better precision of ERP estimation. To verify the above deductions, two experimental schemes are designed and real GNSS data is processed. The results show that 1) the hypothesis of EDOP is correct, where smaller EDOP value corresponds to better precision of ERP solution; 2) the linear functions can be adopted to describe the relation between the ERP solution precision and the EDOP value; 3) the solution precision of pole motion is more sensitive to EDOP variation than that of UT1-UTC.