This research addresses the problem of software compensation of quasistatic machine-tool errors. These errors are estimated to account for about 70% of the total errors of a machine-tool. A method to compensate such errors, therefore, has the potential for bringing about tremendous improvements in the machine's accuracy. Software error compensation approaches are very economical but not very effective because of a lack of proper models for understanding the error generation and propagation process. A goal of this research is to develop a framework for mathematically modeling quasistatic errors. In the actual application of such models in compensation schemes, the model's parameters have to be regularly updated (to account for the slow variations). Preliminary results have indicated that it is possible with a few measurements in the machines workspace. This forms the basis of a practical and cost-effective method of estimating and compensating quasistatic errors to improve the accuracy of machine-tools. Studies to generalize this approach will also be included in this project.