To determine the morphological variation in flowers and its correlation with environment factors and morphological diversity among or within populations, five natural populations of Chimonanthus praecox in Zhejiang Province were chosen as research subjects. We investigated 16 floral phenotypic traits and used nested analysis, principal component analysis, correlation analysis, and un-weighted pair group method with arithmetic averages (UPGMA) cluster analysis to analyze the experimental results. The results showed that there were extremely significant differences among most of the floral characteristics of Chimonanthus praecox. The average variation coefficient of ovule number was highest, and the average variation coefficient of corolla inner diameter was smallest. The corolla inner diameter was the most stable phenotypic trait. The mean coefficient of variation (CV) of all traits was 15.03% with the range from 10.81% to 23.38%, and the average phenotypic differentiation coefficient (V_st) was 44.38% with the range from 1.57% (pistil number) to 89.62% (corolla diameter), phenotypic variation within a population was 55.62%. Average value of CV of the Wujianshan population was the largest and that of the Longwangkan population was the smallest. The variation within populations was greater than that among populations, which indicated that the former was the main source of the floral morphological variation. The principle component analysis showed that corolla diameter, corolla inner diameter, flower cylinder depth, and middle perianth length contributed the most variation. Correlation analysis showed that there were significant and highly significant correlations between parts of floral traits. The soil acidity appeared to be the most prominent ecological factor influencing phenotypic traits of flowers. The results of the analysis on the correlation between floral morphological characters and environment factors indicated that none of the studied environment factors played a significant leading role to affect the variation of one single character. Different environment factors could have different influences on variation of a single character, whereas the same environmental factor showed somewhat different effects on different characters. Macroelements had greater correlation with floral traits than did microelements and mid-elements. According to UPGMA cluster analysis, five populations could be classified into two groups. The UPGMA cluster analysis showed that the morphological variation of the five natural populations of Chimonanthus praecox were not fully clustered in accordance with the geographical distance. It is suggested that there is rich variation and diversity of floral traits among wild Chimonanthus praecox populations, corolla diameter, the length of the middle perianth, and the ratio of length to width of the middle perianth had significant positive correlations with the content of copper, the ovule number had a significant positive correlation with the content of calcium, the ratio of length to width of the middle perianth had a significant positive correlation with the content of iron, and the number of inner perianth had a highly significant positive correlation with altitude of the habitat.