Abstract: The measurement uncertainty of the analytical results of Tl in groundwater samples by inductively coupled plasma-mass spectrometry (ICP-MS) was evaluated using continuous propagation model of uncertainty. The regression of double error was used in the course of standard curve fitting. The extended uncertainty of measurement results was obtained from the synthesis of uncertainties from each source, but the uncertainty from pretreatment procedure was not considered. When the thallium content of the actual samples is relatively low, the uncertainty generated in the course of standard curve fitting is comparatively large and even becomes the dominant factor. When the content of thallium is high, the uncertainty from standard curve fitting is relatively insignificant.