Abstract: In order to solve the problems of the industry standard NY/T 889—2004 for determination of slow-release potassium in soil, including difficult oil-bath heating operation,easy sputtering of nitric acid, and low sensitivity of flame photometer. The oil-bath heating was replaced with the graphite heating,and the digestion tube and funnel in NY/T 889—2004 were replaced with a self-made condensation reflux digestion tube, the study mentioned by title was proposed. 2.50 g of air dried soil sample that passed through a 16 mesh (1 mm) sieve was taken and placed in a self-made condensation reflux digestion tube. 25.0 mL of 1 mol·L⁻¹ nitric acid solution was added and shaken well. The mixture was placed in a graphite digestion apparatus, boiled at 135 ℃ for 12 min (counting from boiling) and poured into a funnel. The filtrate was taken and placed in a 100 mL-volumetric flask. 15 mL of 0.1 mol·L⁻¹ nitric acid solution was used to rinse the digestion tube and the solution was poured into the funnel . This process was repeated 4 times, all filtrates were combined, and the mixture was made its volume with water. The content of slow-release potassium was determinde by inductively coupled plasma atomic emission spectrometry under the conditions of radio-frequency power of 1 150 W, auxiliary gas flow rate of 0.6 L·min⁻¹, and analysis spectral line of 766.491 nm. As shown by the results, linear relationship between the corresponding the response intensities and mass concentrations of slow-release potassium was kept in the range of 0.10-30.00 mg·L⁻¹, with detection limit (3.143s) of 1 mg·kg⁻¹. Test for recovery was made according to the standard addition method, giving results in the range of 88.0%-97.5%. This method was used for the analysis of 15 standard substances, and the determined values were found within the uncertainty range of the certified values, and RSDs (n=7) of the determined values were not more than 7.0%.