Affymetrix SNP arrays are important tools and widely used for both genotyping and copy number estimation. We found previously that systematic bias exists in Affymetrix SNP arrays and leads to loss of accuracy in copy number estimations. In this paper, we showed that systematic bias within SNP arrays is multiplicative effect (or additive effects in the log scale) to probe intensities. Based on this observation, a method was introduced to correct the systematic bias within SNP arrays.