The purpose of this study was to survey chemical distribution of inorganic elements and ions in the submicron particles, to characterize qualitatively emitting sources by factor analysis, and finally to reveal existing patterns in terms of chemical compounds by a stepwise multiple regression analysis. Total of 141 samples were collected by a cascade impactor from 1989 to1996. Fifteen chemical species (Al, Ba, Cd, K, Pb, Cu, Fe, Ni, $Cl^-, NO_3^-, SO_4^{2-}, K^+, Mg^{2+}, Ca^{2+}, and Na^+$) were characterized by AAS and IC. The study showed that average seasonal levels of submicron particulate matters $(d_p<0.43 \mum)$ were 18.7 $\mug/m^3$ in spring, 15.5 $\mug/m^3$ in summer, 15.7 $\mug/m^3$ in fall, and 24.5 $\mug/m^3$ in winter, respectively. All of the anion concentrations in the particle were highest in the winter season. By applying a factor analysis, 5 source patterns were qualitatively obtained, such as sulfate related source, nitrate related source, oil burning source, calcium related source, and coal combustion source. Finally, when applying a stepwise multiple regression analysis, the results clearly showed that $Na^+ and Ca^{2+}, K^+ and Ca^{2+}, NO_3^-$ and relative humidity, $Cl^-$ and ambient temperature, $Ca^{2+} and Cl^-, Mg^{2+} and SO_4^{2-}, Na^+ and NO_3^-, and Ca^{2+} and NO_3^-$, respectively, are negatively contributed to each other. As a result of those statistical analysis, we could suggest that some chemical compounds in the submicron particles such as$NaNO_3, MgSO_4, Ca(NO_3)_2, and CaCl_2$ may not exist on the filter as final composing products; however, other compounds may possibly exist in the form of $Mg(NO_3)_2, CaSO_4, Na_2SO_4, K_2SO_4, MgCl_2, NaCl, and KCl$. Thus, it must be necessary to identify differences between the results of above statistical analysis and of the real world by laboratory experiments.