In ultra-precision machine tool, the running accuracy of aerostatic bearings has a great influence on the machined surface topographies. However, the mechanism of aerostatic bearings running accuracy has not been fully understood. In this paper, a method based on computational fluid dynamic (CFD) method and dynamic mesh technology (DMT) was proposed to quantitatively study the effects of manufacturing errors on the running accuracy of aerostatic porous bearings. The different types of waviness errors and non-flatness errors are modeled based on the actual measurement results of spindle and thrust bearing, respectively. The DMT was applied in CFD method to simultaneously solve the Navier-Stokes (N-S) equations and the Newton's law. The calculation results show that the radial running accuracy of journal bearing can be improved by reducing the waviness amplitude or spatial wavelength, the axial running accuracy of thrust bearing increased with the decrease of non-flatness amplitude. Besides, a nanometer system for measuring the running accuracy of aerostatic porous bearings was constructed based on the Donaldson reversal method. The bearing rotation movement trajectory of calculation results was very similar with the experiment results, which verified the validity of calculation method proposed in this study. Both the calculation results and experimental data confirmed that the effect of waviness errors on the bearing running accuracy was much more obvious than non-flatness errors, which provide the useful guidance for the design and manufacturing of aerostatic porous bearings.