It is very important to studying the genetic diversity of sea island cotton based on genetic and phenotypic characteristics, which can provide a theoretical basis for sea island cotton hybrid breeding in selecting parents, introduce new germplasm resources for broadening the range of genetics and breed new varieties. This study analyzed the genetic diversity of Sea Island cotton in 94 germplasm based on the identification of 125 SSR molecular markers and investigation of 13 phenotypic traits in the field in two years, and classified into different groups and comparied them according to their differences in genetic traits and phenotypic traits. The results were as following: (1) there were 420 loci being detected using molecular markers, of which 249 loci were polymorphism, and Nei-Li similarity coefficient method were applied to estimate the genetic similarity coefficient between 94 sea island cotton varieties. It was showed that the genetic similarity coefficient was between 0.46~ 0.95. Meanwhile these 94 sea island cotton varieties were divided into four groups using SSR molecular markers and the classification results were in agreement with the results of pedigree analysis. (2) The results of phenotypictraits showed that the range of the quality traits were wide while the range of yield traits were narrow. According to the diversity cluster analysis, the diversity index of micronaire value was the highest among these traits, on the contrary, the diversity index of single boll weight was the lowest. It was also found that type Ⅰ had wider range in yield and the quality in the middle of these types; type II was with low value both in yield and quality; type III had higher mean value in yield than others; and type IV was not good at yield trait, but it’s fiber trait was better than others. (3) Combination of SSR markers and agronomic traits analysis showed that the genetic diversity of 94 sea island cotton germplams was rich. These two methods had the same results in classification and the results of clasification had been related with geographic distribution.