Abstract: A method was developed for the rapid transformation of thiosultap-disodium into nereistoxin in plums, combined with HPLC-MS /MS for the determination of nereistoxin and the determination of thiosultap-disodium residue by the content of nereistoxin. The samples were extracted with 0.1 mol/L hydrochloric acid solution containing 1% L-cysteine hydrochloride, derived with ammonia water and nickel chloride, and then extracted with n-hexane. The volume was constant with methanol after N₂ drying. HPLC-MS/MS was used to detect and quantify the residues by external standard method. The results showed that the conversion rates of thiosultap-disodium to nereistoxin in plum substrate were decreasing with the increase of concentration. In the range of 0.05-10 mg/kg, there was a power function correlation between the concentration of thiosultap-disodium and the peak area of nereistoxin, and R² was 0.992. The average recoveries were 77%-100%, and the relative standard deviations were 6%-18% at three spiking levels of 0.01, 0.1 and 5 mg/kg. The field trial results showed that after the 18% water aqua of thiosultap-disodium was sprayed at 360 and 540 mg/kg for 2-3 times on plum trees, the residues of thiosultap-disodium in plum was <0.010-1.4 mg/kg at 10, 15 and 20 d after the last application. Combined with the pesticide registration status in China and the per capita dietary structure of Chinese residents, STMR, HR and MRL values were used to calculate the proportion of thiosultap-disodium residues in plums for the long-term dietary exposure risk. The results indicate that the dietary intakes calculated by STMR, HR and MRL were 0.17, 0.70 and 0.80 mg, respectively. The intake calculated by HR and MRL exceeded the daily allowable intake (0.63 mg) of thiosultap-disodium. The corresponding risk portion was 0.89%, 4.93% and 0.76%, respectively. It can be seen that the risk uncertainty would be greatly increased by using HR or MRL to calculate the dietary intake of thiosultap-disodium in plums. Therefore, it should be prudent to use HR or MRL values instead of STMR value in the process of dietary risk assessment.