LA-ICP-MS was used to calibrate 26 turquoise samples with relatively clean and uniform color on each surface. We selected 9 points on each sample to test the homogeneity of their chemical composition. The results showed that the average error coefficients of Al, P, K, Cu, Fe, V, Cr and Zn in turquoise samples are below 5.4%, while the average error coefficients of CaO and SiO2 are 34.8% and 16.2% respectively, suggesting that the elements of Si and Ca in turquoise were heterogeneity. The turquoise working curve can be established by the energy dispersive X-ray fluorescence spectrometer (EDXRF). We used 21 samples as the standard sample for reference and 5 as unknown samples to determine the composition of turquoise. Results showed that the correlation coefficients of Al, P and Cu of the main element are between 92.3% and 94.3%, with an average relative error of 4.6%~9.7%. The correlation coefficients of Fe, Cr, Zn micro-elements are more than 0.990, while the correlation coefficients of both K and V elements are 0.939 and 0.972. And the average relative error of those five micro-elements ranges from 7.2% to 13.9%. However, the correlation coefficients of Si and Ca are 0.958 and 0.866, the average relative error of which is 348% and 27.8% in the five unknown turquoise samples. The high average relative errors of Si and Ca may be influenced by the low content and the heterogeneity of Si and Ca in turquoise, also related to the detection limit of this instrument method. The repeatability test showed that the relative standard deviations (RSDs) of Al, P, Cu, Fe, Zn elements are within 1%, suggesting that the accuracy of the test results is respectively high. While the RSDs of V, Cr, K, Ca, Si elements range from 1.34% to 10.17%. The study provides a new idea and method for determining the quantify of the elements of Al, P, Cu, Fe, Cr, Zn and V in turquoise quickly, accurately and losslessly, which can be applied to test and identify turquoise in laboratory.