On-Machine Measurement (OMM), which measures workpieces where machining is progressing or finished in CNC machining center, has its own strength that is capable of measuring workpieces directly in work space without moving them. However, planning of measurement to determine measurement sequence and elements for each shape of complicated objects has its limitation requiring time-consuming tasks to generate measurement points mostly relying on skills of workers on site. This paper will suggest how to apply an existing and highly similar model’s measurement paths to a new model by analyzing similarity between existing 3D shapes whose measurement path are built and new models. For the purposes of similarity analysis, this paper extracted feature vectors from 3D shapes models which are able to express characteristics of shape models and then applied this feature vectors to histogram of probability distribution based on algorithm of similarity analysis. In addition, this paper developed CAIP system by correcting unapplied measurement points, which are generated by minute difference between two models when their similar measurement points are applied to a new model and then generate final measurement paths.