In the multi-GNSS (global navigation satellite system) era, the number of navigation satellites from GPS, GLONASS, Galileo, BDS, QZSS, IRNSS, and so forth is more than one hundred (Kaplan & Hegarty, 2017;Morton et al., 2020). Yet, if all visible satellites are used in the navigation processing, the demands on the computational load and power consumption are high while improvement on performance is marginal. Thus, in practice, it is desired to investigate the issue of GNSS satellite selection with the aim of selecting a subset of satellites for navigation processing with little degradation on performance. The paper aims to provide a general formulation of the performance-based satellite selection problem and propose a numerically feasible method for the selection of satellites.Many satellite selection methods have been investigated in the past. It is known that high elevation satellites are less susceptible to errors due to atmosphere and multipath, one may thus rank the satellites in terms of their elevations and pick some high elevation satellites for navigation. In contrast, the geometric distribution of satellites is another important factor to be considered in affecting the