A method of precision evaluation for field large-scale measurement instruments is proposed with multi-instrument stations and multi-reference points. The measured data from multi-instrument stations are unified in the global coordinate system using spatial data registration based on quaternion. The measurement uncertainty is obtained by statistics of measurement values based on controlling points coordinate invariable constraint, so that the uncertainty of each sensor component is acquired. The optimal controlling points are calculated by multi-sensors information fusion technique, and the effect of coordinate transform error is decreased. Taking laser tracker as an example, the simulation and field experiments show that the uncertainty error of range and angle are decreased to 1 μm and 0.1″, respectively.