Abstract: 29 varieties of Camellia oleifera were selected, which widely cultivated and representative in Jiangxi Province. 10 phenotype traits were analyzed statistically, meanwhile phenotypic diversity level and genetic relationship were investigated based on repeatability, Shannon-Wiener diversity index and clustering results, using variance analysis, correlation analysis, cluster analysis and principal component analysis. The results showed that a wide phenotypic variation in trees, leaves and fruits was presented, the average coefficient of variation was 11.09%, the average relative extreme value was 38.08%, and the average phenotypic diversity index was 2.356 5. Repeatability of fruit traits (0.99) and leaf traits (0.95) was significantly higher than the tree traits (0.71). Choosing fruit traits in breeding will provide a higher genetic gain under the same selection intensity. Phenotypic traits have a significant correlation between the output rate of seed and oil content. These 29 different varieties were divided into 5 categories by cluster analysis and principal component analysis. Phenotypic diversity level among varieties has higher differences.The fruit and leaf traits phenotypic diversity difference caused by the genetic differences, in addition, environmental conditions had a greater influence on tree traits. Selecting important index, which significantly related to yield, can forecasted production. Cluster analysis and principal component analysis results were basically identical, both could able to distinguish the genetic relationship between different varieties. Most accessions from the same origins of varieties has similar phenotype, which may indicates that the genetic distance closer.