A method for the determination of talc power in soap by inductively coupled plasma atomic emission spectrometry (ICP-AES) with graphite digestion tube was proposed. Two soap samples (0.5 g each) treated were put into two graphite digestion tubes, and 10.0 mL of nitric acid and 2.0 mL of 30% (w, the same below) hydrogen peroxide solution were added into one tube for determination of impurity magnesium except for magnesium silicate, and 10.0 mL of nitric acid, 2.0 mL of 30% hydrogen peroxide solution and 2.0 mL of hydrofluoric acid were added into the other tube for determination of total magnesium. The open tubes were placed for 1 h, and then the mixtures were digested at 120 ℃ for 2 h. After digestion, the solutions were cooled to room temperature, and their volumes were made up to 100 mL with 0.5 mol·L^-1 nitric acid solution. Magnesium was determined respectively according to the working conditions of the instrument, and the content of talc power in the sample was calculated by the difference between them, and the working curve was established with matrix matching method. The results showed that the linear range of magnesium working curve was 0.100-3.00 mg·L^-1, with detection limit of 0.002 g·kg^-1. RSDs (n=6) of the determined values of talc power in the actual sample were in the range of 0.80%-2.7%, and the spiked recoveries were ranged from 97.3% to 109%. Three soap samples were analyzed by flame atomic absorption spectrometry in GB 5009.269-2016 and method. At 95% of confidence level, the t values of the three samples were lower than t_0.05(4), indicating that there was no significant difference between the two methods, while the detection limit of this method was lower and the precision was better.