AbstractCalibration methods that apply relative techniques to map robot inaccuracy increase robot flexibility by allowing for greater use and application of off‐line path planning techniques. To successfully implement the off‐line techniques described in this article, methods have been integrated for off‐line tool control frame (TCF) calibration, data base integration, and on‐line robot inaccuracy correction. The TCF calibration methods presented in this article successfully determine the TCF origin of the end‐effectors and sensors attached to the robot distal link to within the robot repeatability. When the methods were applied to a SEIKO DTRAN RT‐3000 robot, the TCP origins were statically predicted correct to within 0.025mm (0.001 in.). On‐line methods determine robot inaccuracy by mapping robot configurations relative to operational parts or to spatial templates, in contrast to the contemporary global mapping approaches. For the Seiko robot the resulting inaccuracy distributions permit simple error correction equations to be applied, improving robot positioning accuracy to less than ±0.1 mm in locally large regions and to within the repeatability of the robot ±0.025 mm) in smaller regions. This represents an order‐of‐magnitude decrease in robot inaccuracy.