The purpose of this study is to apply different methods including statistical and spatial analysis techniques to delineate spatial submarkets of housing prices and to examine spatial dependence of housing prices. The data comes from housing transaction prices in the central development areas of greater Tainan City, 2009. Greater Tainan city is a new metropolis amalgamated from former Tainan city and Tainan county. Due to the amalgamation of municipalities, the local government boundaries should be adjusted, and in the mean time, it is worthy to identify spatial submarkets of housing prices in greater Tainan City, compared to former political boundary submarkets. It was found that spatial submarkets of housing prices classified by cluster analysis and spatial techniques are similar. Higher housing prices are concentrated in the core of central development areas while lower prices spread widely around outer ring of the central development areas in greater Tainan city. In testing spatial autocorrelation of housing prices, it was found that it exists significant spatial dependence between housing prices. In modeling housing prices, the results show that spatial submarkets derived by spatial autocorrelation techniques have stronger and higher impacts on housing prices, and the model also have better goodness-of-fit compared to other two types of models.