The objective of this study was to select the elements associated with the geographical origin of rice and evaluate the traceability efficiency of these elements by a chemometrics method. The contents of 15 elements (Mg, K, Ca, Na, Be, Mn, Ni, Cu, Cd, Fe, Al, Cr, Zn, Sb, and Pb) in rice samples and the contents of their exchangeable, available form, and "total" form in soil samples from four provinces of China were analyzed by inductively coupled plasma mass spectrometry and flame atomic absorption spectrometry. The results of variance analysis and correlation analysis demonstrated that there is a significant difference in the contents for 9 of 15 elements (Mg, K, Ca, Na, Be, Mn, Ni, Cu, and Cd) in the rice samples from 4 provinces, and the 9 elements in rice samples are closely connected with the mineral elements in soil. R-type hierarchical cluster analysis of 15 elements in rice indicated that Mg, Cu, K, Ni, Be, Mn, and Ca clustered together with similar behavior in rice and soil. The predictions of the geographic origin made by linear discriminant analysis (LDA) based on 9 elements gave an overall correct classification rate of 100.0% and a cross-validation rate of 93.8%, which was superior to the result for the total 15 elements. The correct rate of Q-type hierarchical cluster analysis (Q-type CA) based on 9 elements was higher than that based on 15 elements. The result of Q-type CA and LDA demonstrated that the 9 elements are good indicators to discriminate the geographical origin of rice.
