The invention discloses a method for accurately estimating end-point-method reagent blank absorbance. The method comprises the following steps: dividing the absorbance of a solution subjected to a reaction into four parts, namely residual reagent absorbance, consumed reagent absorbance, product absorbance and sample residue absorbance; primarily estimating a reagent blank by selecting a specific secondary wavelength and estimating a sample concentration; subtracting the consumed reagent absorbance, the product absorbance and the sample residue absorbance by the absorbance of the solution subjected to the reaction; estimating again to obtain the accurate reagent blank. By adopting the technical scheme, the reagent blank does not need to be tested actually; a principle is clear and a calculated result is accurate and reliable; the estimation error is less than 1%; therefore, the method is especially applicable to an instrument which cannot directly measure the reagent blank.