Introduction: In SPECT, usually a large number of scattered photons are detected.Therefore the reconstructed image without scatter compensation has degraded image quality and biased quantization. While the efforts made to compensate the scatter effect, none of them can perform fast and accurate scatter compensation in non-uniform scattering objects.Methods: and materials: A class of scatter compensation methods, called reconstruction-based scatter compensation method, RBSC, is based on modeling scatter effects in the transition matrix used in iterative reconstruction methods. The accuracy of this method is dependent upon the accuracy of scatter model used.Beekman et al 1997 have shown that RBSC methods results in images with less variance when compared with subtraction-based scatter compensation methods. The main disadvantage of RBSC methods is that the scatter models tend to be very computationally intensive. In this paper we would present a mathematical approach for further reducing the calculations and time of reconstructions using subtraction in the iterative reconstruction methods. In this algorithm scattering contributions of each pixel, from activity of 27 neighbor pixels in 3 slices, which are along the detector is estimated for all detector bins, using Klein-Nishina formula. These data are stored in a certain file and can be used in all iterations in RBSC process. The iterations start with an uncorrected image which is estimated using MLEM formula and then it is corrected subsequently for scattering using: Where fj represents one pixel in the image space, gi is the measured SPECT emission data with detector, and ai j is the coefficient that represents contribution of image pixel j to detector i. Index l denotes pixels number j in projection bin i, so that summation over l makes the projector, and the summation over i the backprojector. SCl is the scattering contributions of pixel j from neighbors\' pixels on slice s to detector i, which is calculated from K-N formula as below: Results: The result on heart phantom and patient images showed that the proposed algorithm significantly improves the contrast and resolution of images. The RAM of the computer and time of reconstruction is dependent to the image size.Conclusions: The proposed algorithm effectively compensates scattering effects in SPECT images and is capable to modify clinical images using a pc in routine clinical activity.