The aim of this study is to analyze the effect the factor of urban characteristics on the amount of damage caused by natural disaster using panel data which is composed of time series data of 1998 through 2010 and cross section data of 20 Cities in Gyeongsangnam-do province. Specifically, this study considered the urban characteristics which have been ignored Previous related studies. The element of urban characteristics which composed ten variables and two dependent variables about the extent of the damage caused by natural hazard are selected for this analysis and five models were used to find the optimum model including OLS and TSCSREG(Time Series Cross Section Regression). The result proved that the most suitable model is random effect model of One-Way Error Component Regression Model. Each model in this empirical analysis is as described below. The urbanization density and city budget are positively significant variables. Also, the gross area and the ratio of the housing are negatively significant variables. Therefore, cities will set up an effective policy to prepare for the natural disaster.