Tallium in the blood of occupationally exposed population may reflect their exposure in vivo. Therefore, it is very important to establish a method to determine thallium in blood. At present, there is no national standard method for the determination of thallium in blood in China, and the methods reported in domestic and foreign literature have some shortcomings. In order to obtain accurate thallium concentration in the blood of occupationally exposed population, a new method was developed for thallium in blood by inductively coupled plasma mass spectrometry with high matrix injection. In this study, plasma mode and pretreatment method were selected. After the 0.20 mL sample was pretreated with the mixed solution of 0.1% Triton X-100 and 0.5% nitric acid, the thallium content was detected by adding internal standard online. Under the optimal conditions, good linearity was obtained in the concentration range of 0.02~4.00 μg·L^-1 with a linear calibration curve of Y=0.010 33X+0.000 12 (R=0.999 9). The detection limit (DL) and the quantitative limit (QL) were 0.005 and 0.02 μg·L^-1. When the sample volume and constant volumes were 0.20 and 5.00 mL respectively, the method detection limit (MDL) and the method quantitative limit (MQL) were 0.12 and 0.42 μg·L^-1, and the analytical range of this method was 0.42~100 μg·L^-1. The average recoveries at three fortified levels (2.50, 15.0 and 75.0 μg·L^-1) were 92.7%~103.8%. The relative standard deviations of in-batch (RSDs of in-batch) ranged from 1.71% to 2.81%, and the relative standard deviations of interbatch ranged from 2.84% to 4.77%, which indicated the accuracy precision of the method were good. The signal variation of ²⁰⁹Bi internal standard isotope in 50 samples(including standard solution, quality control samples and whole blood samples) was less than 7.7%, which showed that the method had good stability. The established method was successfully applied to the analysis of 30 whole blood samples in the occupational exposed population. Four samples were above the MDL, but below the MQL. The other 26 samples were below the MDL and all the 30 samples were within the average background range. This method was simple, rapid, accurate and stable and suitable for analysing of large quantities of actual samples. The results showed that the exposure level of thallium was shallow in vivo, and there was no potential health risk of thallium to 30 professional contacts in their workplace.