Gas Electron Multiplier (GEM) based coordinate detectors are used at different high energy physics centres and at Budker Institute of Nuclear Physics particularly. These detectors possess a spatial resolution in ten micron scale together with high rate capability up to 107 cm−2s−1. Thus, the precise investigation of best possible spatial resolution, achieved with GEM-detectors, is the subject of interest. The experimental data, accumulated by the moment, gives the possibility to compare it with the simulation results. The simulation of applied detector configurations includes transport of electrons through the detector and tracking of avalanche evolution inside the working volume, as well as obtaining signal distribution on the readout strips. The spatial resolution, obtained in the simulation of an individual detector, is found to be essentially better (the difference is about two standard deviations) than the experimental results. Further efforts to find out the reasons of the contradiction between the simulation and measurements were made. In particular, the simulation of complete experimental set-up (including tracking detectors) was performed. The results of individual detector simulation and the simulation of complete set-up were determined to generally coincide.