To determine the phenotypic variation in Rosa rugosa populations and the relationships between phenotypic variation and different distribution areas, 24 phenotypic traits from 5 representative populations were chosen as research subjects. Analysis of variance, multi-comparison, principal component analysis, correlation analysis, and cluster analysis were used to analyze these data. The results showed the following. (1) There were highly significant differences in the 24 phenotypic traits among and within the populations, and the variation was high. The mean phenotypic differentiation coefficient (V_ST) among populations was 0.2952, which indicated the phenotypic variation was great. The variation within populations (27.18%) was greater than that among populations (16.79%), which indicated that the former was the main source of the phenotypic variation in the species. The mean phenotypic variation coefficient (CV) was 18.48%, and the variation range of the 24 phenotypic traits was 6.67%-26.79%. The variation coefficients of leaves, fruits, flowers, and seeds were 21.40%, 17.42%, 12.54% and 6.67% respectively, which indicated that seeds exhibited the greatest stability, whereas leaves exhibited the least. The principle component analysis also showed that the phenotypic variation of leaves and fruits contributed the most variation. (2) The correlation analysis showed that there was significant correlation between some traits and geographic ecological factors. With the decrease of latitude and longitude, the length of stipule decreased, but when annual temperature and the average July temperature increased, the length of stipule became even greater. Increases in annual precipitation resulted in larger fruit diameters, and the ratio of the transverse and longitudinal diameter of fruits and also increased, which indicated that the shape of fruit varied from broadly orbicular to oblong. A decrease in average July temperature and an increase in latitude and longitude resulted in an increase in thousand-grain weight. (3) The five populations of R. rugosa investigated could be divided into three groups according to the UPGMA cluster analysis; the phenotypic traits gathered were based on geographic distance, and it was noteworthy that the population of Chengshan town in Shandong Province and the population of Mingyang Town in Liaoning Province were in the same group.