Coordinate Measuring Machines (CMM) are widely used in industry, mainly for assessing product’s compliance to geometrical specifications. They are a de-facto standard for such a control, owing to their high accuracy and flexibility in catering to different measurement tasks with fast and easy setup change. One of the key aspects of any CMM inspection plan, affecting accuracy and cost, is the choice of the measurement strategy, namely the definition of the number and the location of the points sampled on the part’s surface. In the industrial practice, such important decisions are usually left to the experience of qualified operators. A method is presented here, based upon an innovative application of Kriging models to coordinate metrology, that defines an adaptive sampling strategy with the goal of obtaining accurate estimates of form errors using relatively small samples. Exploitation of kriging models (originated for geostatistics investigations, then successfully applied to the sector of Computer Experiments) in dimensional metrology entails however some adjustments. Two practical cases of form error estimation have been investigated, the first dealing with flatness verification, and the second with roundness.