Quasistatic errors are the major contributors to positioning inaccuracies of machine tools. These errors, estimated to account for 70 percent of the errors of the machine, have been observed to be as high as 70 to 120 μm for production class machining centers. In this paper, a model for these errors is developed. Further, it is shown that the parameters for this model can be estimated by the observation of a few (nine) error vectors in the machine’s workspace. These two results form the basis for a viable error compensation scheme; the model for error compensation and the parameter estimation for updating the model. Experimental results verify the validity of the approach. An order-of-magnitude improvement in the accuracy of the machine, with respect to a calibration frame, was observed.