The aging state of transformer insulation papers was not assessed accurately. In this paper, multi-feature comprehensive assessment method was proposed to convert the nonlinear problem into a linear problem fitting for assessing the aging state of insulation papers. The whole aging process was divided into two stages with the degree of polymerization of 500 as the dividing point. The backward multiple linear regression analysis model was established and the error of the fitting results was analyzed. Most of the relative assessment errors were within ± 5 %. It compensated for the deficiency of aging assessment of transformer insulation paper based on single feature, and the accuracy of assessment results was improved.