The present paper analysed in detail the reasons for the errors found in the literature when Boltzmann plot was used for the temperature measurement of a high-density plasma produced by electrothermal capillary discharge. The reliability of experiment can reach as high as 98%-99.5% and the error is only +/- 6.5% when the spectroscopic parameters, such as transition probability A, statistical weight of upper level g, and the energy of upper level Ei, are selected correctly for the temperature measurement of a high-density plasma produced by electrothermal capillary discharge by Boltzmann plot in atomic emission spectroscopy.