To improve the measurement accuracy of the articulated coordinate measuring arm (ACMA), the measurement system’s parameters must be calibrated. The mathematical model and the disturbance of the kinematic parameters’ error are critical to the calibration precision. In this paper, product-of-exponential(POE) model is adopted to establish the kinematic model of the ACMA, and an error correction model which utilizes single-point versus multi-pose and the distance between every two points is used to take accuracy and repeatability into account. Then, genetic algorithm is utilized to optimize the error correction model in order to identify kinematic parameters. Experimental results reveal that the proposed calibration method is feasible.