We use the Gauss-Newton method to find the parameters in the parameters calibration for PCMM (Portable Coordinate Measuring Machine).Then we find the relationship between the identification of parameters and the conditions numbers of Jacobin Matrix,and in practical the conditions numbers of Jacobin Matrix depends on the sample point.Therefore,the sampling strategy must be optimized.Using the cone and standard bar as a calibration standard,this paper presents an optimized sampling strategy to increase the value range of each joint and thus minimize the conditions number of Jacobin Matrix.Experiment results show that the sampling strategy of calibration greatly improves the measurement accuracy.