Coordinate detectors based on Gas Electron Multipliers (GEM) are used in experiments at many high energy physics centers and at Budker Institute of Nuclear Physics particularly. Spatial resolution of these detectors is known to be in tens microns scale. Also the detectors possess a rate capability up to 107 cm-2s-1. Consequently, the precise study of best possible coordinate resolution, achieved with GEM-detectors, is a significant task. The experimental data, collected by the moment, provides the possibility to compare it with the results of the simulation. The simulation of the detector performance includes transport of electrons through the detector medium, tracking of an avalanche evolution inside the working volume, as well as registering of the signal distribution on the readout strips. The simulation of individual detector shows that its spatial resolution is considerably better than the experimental results with the difference about two standard deviations. In order to find out possible reasons of the contradiction between measurements and the simulation of the individual detector, the simulation of complete experimental set-up (including tracking detectors) is performed. The results of complete set-up and individual detector simulations are determined to coincide in general.