Errors of structural parameters are one of the main factors causing uncertainty of Articulated Arm Coordinate Measuring Machines (AACMM), however, the accuracy of AACMMs can be improved efficiently by identifying structural parameters accurately. In this paper,the homogeneous transformation equations of 6-DOF AACMMs were established based on Denavit-Hartenberg model and a method to identify structural parameters with a cone-shape hole was analyzed. Then,a Modified Simulated Annealing Algorithm (MSAA) was presented to identify the structural parameters of the AACMMs. The MSAA can keep the best solution during search and can reduce the search range when the solution approaches to the global minimum, which improves the search efficiency and accuracy. The definition of single-point repeatability was deduced, and the structural parameters of AACMMs were identified by using single-point repeatability as the objective function based on the MSAA. A 6-DOF AACMM was used to carry out experimental tests to evaluate the efficiency of the presented method, the single-point results show that after identification, the repeatability and length measurement accuracy of the AACMM are improved by 7.87 and 5.59 times respectively.