Abstract: It is difficult to select sampling point exactly when the model based parameter estimation(MBPE) method was used to interpolate the monostatic radar cross section(RCS) by the objects in the wide frequency span and wide angle span, especially when the curve shapes of the monostatic RCS fluctuate intensively. Although the adaptive method can solve this problem, it needs to get the exact numerical value of the relative points in advance and it is difficult to be parallelized. By contrast, the uniform sampling method is easy to parallelize, but the sampling scheme should be known at first. In this paper, based on the computation of the monostatic RCS by the objects, the sampling point selections schemes for MBPE interpolation in spatial domain and in frequency domain were proposed. In spatial domain, the sampling points were located along with the sphere wave directions. In frequency domain, depending on the similarity of the RCS's characteristics by the same objects, the sampling experiments were firstly detected in the low frequency, and then were used to calculate the monostatic RCS in the high frequency.Numerical experiments show that in spatial domain, the sampling points of MBPE are only half of sampling points of the fast multipole method(FMM) and will decrease with the narrowing of the angle's scope; in frequency domain, the amount of the MBPE sampling is independent of the maximum frequency, only slowly increases with increasing sweep range.