The Quantitative Ion Character-Activity Relationship (QICAR) model between metal ionic properties and toxicity order numbers (TON) was established assisted with cluster analysis to find out typical metal ionic properties. Means while, the multicollinearity among the typical metal ionic properties was minimized by principal component analysis during the modeling and obtaining .the QICAR model by multiple linear regression analysis as follows: TON = -70.675AR/AW - 0.353|lgKOH| - 28.312σp - 0.537ΔE0+ 2.436Xm 2r+ 0.240AN - 0.024Z*2/r+ 0.475N + 26.033. It has satisfactory predicted ability with Nash-Suttcliffe Simulation Efficiency Coefficient (NSC) of 0.91 and 0.85formodel building and testing respectively. The metal ionic propertiesXm 2r, AN and N have positive effects on the model, while the metal ionic properties AR/AW, |lgKOH|, σp, ΔE0 and Z*2/r negatively influence the model. Finally, sensitivity analysis of the metal ionic properties in QICAR model was carried out to show which metal ionic properties have more impact on TON, indicating different impact degree of the metal ionic properties on each metal ion.