Abstract: A method was used for determination of 38 inorganic elements in goji fruit from different regions, including 21 conventional elements （Na, Mg, Al, K, Ca, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Ag, Cd, Ba, Li, Pb, Sr, and U） and 17 rare earth elements （Sc, Y, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, and Th）, by inductively coupled plasma mass spectrometry （ICP-MS）. A characteristic chart of inorganic element content in goji fruit was established, and the determination results were systematically analyzed by chemometric methods of principal component analysis, cluster analysis, and discriminant analysis. As shown by the results, the inorganic element characteristic chart of goji fruit from different regions exhibited a relatively regular distribution pattern. Among the 4 different regions, K had the highest mass fraction of 38 elements, followed by Na, Mg, Ca, Fe, Al, Zn, Mn, Cu, and Sr, while mass fractions of 17 rare earth elements were all lower. There were significant differences （p＜0.05） in mass fractions of 17 rare earth elements, Mg, Al, K, Mn, Fe, Co, Ni, Cu, Cd, Ba, and Pb, in goji fruit from Yumen region in Gansu Province compared to Jingyuan region, and there were also significant differences （p＜0.05） in mass fractions of Na, Mg, Ca, Co, Ni, Cu, Zn, Ag, Cd, Pb, Li, and U in goji fruit from Zhongning region in Ningxia Province compared to Qinghai Province region by one-way analysis of variance using IBM SPSS 24.0 software. According to principal component analysis, Al, Fe, Ni, Cu, Zn, Ca, Li, As, Sc, Y, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, and Th were the characteristic inorganic elements of goji fruit. Through cluster analysis and discriminant analysis, the effective indicators including Na, Al, K, Cr, Co, Ni, Cu, Zn, Cd, Pb, and Sc were selected. The variance analysis of the discriminant results showed significant inter-class differences （p＜0.01）, indicating the validity of the discriminant results. The orthogonal validation method was used for verification of the discriminant model, and the accuracy of the regression for 61 batches of goji fruit samples was 100%, with a cross validation success rate of 98.4%, indicating that the obtained discriminant equation was ideal and effective, to provide a method and basis for tracing the region of genuine goji fruit medicinal materials.