Based on 1980～1987 temperature, pressure and humidity soundings measured inNanjing and its surrou nd ings an expression is formulated for the vertical d istribution of atmospheric refractivity index and used to calculate the distribution parameters, fitting errors and Nunit vertical change gradients at a range of climate intervals, Results show that the annual meanvertical parameter differs slightly at these intervals but they are different considerably one fromanother on a yearly mean basis; The vertical distribution parameters differ from each other greatly on a monthly mean basis for each year, showing pronounced seasonality. The distribution ofthe monthly mean parameters in each of the years gives a clear picture of characteristics of climatic intervals, the interannual seasonal and intermonthly characters. Finallv,a muItivariant statistical prediction model is constructed with random vibration phases of a time series considered andused to make forecasts of the vertical distribution of monthly mean parameters for each of the vears.