Abstract: Canonical correlation analysis (CCA) is a multivariation analysis technique examining the maximum correlation between two sets of variates. This paper used it in studying the correlation between the plant importance values and soil components in plant communities, also compared it with PCA. The results indicated: CCA is very useful in explaning quantitatively the correlation between two sets of variates, but also showing the complex-action of many factors. However, the collected data are needed standarded and PCA analysing before applicating this technique.