Abstract : There is an economical challenge in finishing operations of metal and composite parts, which are faced with relocation. Robotic machining turns to be a good solution for machinists to keep their job. As a matter of fact, industrial robots are getting cheaper while their performances are getting better. However their stiffness is still not sufficient to perform finishing operations that require both high precision and high accuracy. Therefore, this doctorate thesis aims at developing and validating methodologies to control and optimize the use of industrial robots for metal or composite parts trimming. Chapter I deals with the main issues in robotic machining and a literature of the research works done in that field. The sources of flexibility of six-revolute serial robots are analyzed in Chapter II in order to use an elasto-static model suitable for the problem at hand. A methodology for the identification of the joint stiffness values is developed and applied in Chapter III. Then this elasto-static model is validated experimentally by means of static tests and some machining operations. In Chapter IV, the elasto-dynamic behavior of the robot is studied analytically and experimentally. Finally, two methods to realize trimming operations with an industrial serial robot are presented in Chapter V: (i) the workpiece placement optimization and the control of the kinematic redundancy of the robot due to the tool rotation; (ii) the correction of the reference trajectory of the robot. Those methods are validated experimentally.