(1) Purpose e scarcely find sue the case that all of thecities in a country have the same population size. Questions concerning the existence of such size variance in a set of cities originated this study. At present, The city size distributions are distinguished into three types of model; the rank-size rule as Zipf model, Christaller`s hierarchical size groupings and Jefferson`s law of primate city. The primary objective of the present research is to identify what type of model fit good this country. Especially this paper attempt to test the Zipf model as a measure of th?stability of the overall size distribution of cities over time in South Korea. The logic of the above attempt is based on the systems theory which describes the stability of city systems as a state of equilibrium in which entropy has bean maximized. (2) Data organization Two sets of population data are compiled. Each set is composed of three series of selected years. The First set is for the upper 40 cities for the whole country in 1921, 1929 and 1935 before emancipation and the second is for the citie: with populations over 50,000 for the Republic of Korea in 1960. 1966 and 1975. Population size for each selected year was plotted against its rank order on a double logarithmic paper and then the regression equation was formulated. The deviation of the actual population size against the expected was measured by calculating several deviation indices. Lastly, changes in the rank order of 45 cities for 4 selected years were plotted on a semi-logarithmic paper. And Kendall`s coefficient of concordance for those years was calculated for the sake of revealing the over all reationships among rank orders. (3) Implications and conclusion Neither hierarchical groupings of places nor complete rank-size distributions were found on the curves. Dstinctive contrast can be found for the overall shapes of distribution curves between two matrices. For the former, on the one hand, the curves trend downwards with several irregular steps especially in 1921 and 193, on the other hand, the curves for the second set of data conform more or less to straight line distributions. It seems that simple agricultural environment in 1920`s and rapid industrialization in 1930`s cause the two irregular size dis tribution respectively. The regression coefficients are less than unity before emancipation. After emancipation, that values of regression slopes are larger than unity. The parameters q and P, have been increasing continua 113 through the all series of years. It indicates overgrowing population size of the capital city. .Neither curve confearms to the rank-size model completely. In each case the deviation of the capital city from the model is too big. The continuing concentration of population into capital city decribces the deviation amplifying process in the city systee. In this context the stability of the system of cities is questionable. The changes of rank order within system of cities are small for the upper part of the size array, but considerable for the lower part of the curve. The unstability of the rank order for the lower part of the curve is an indicator of the changes in socio-economic environment. Answers to the existence of stability in the city s,-stems need more long term investigation for deriving more desirable result. Despite of the great divergency of the actual size against the expected one, the rank-size model constitutes a meaningful tool as a rneasure of stebilization process of the overall size distribution and of divergences from the probable city sues.