In this paper we estimate relative consumer price levels as of 2008 for 36 major Chinese cities, using an innovative method purposely designed to rectify three main defects of the existing literature, which are (1) the under-representation of marketized services in the sample data, (2) biased consumption weights, and (3) a mismatch between sample classification and consumption weights. Our estimation results show the “subnational Penn effect” as defined by Tang (2012), i.e., strong inter-city correlations among population size, the relative price level, per capita nominal and real income, and human capital stock, thereby showing that the theoretical model of inter-city price dispersion proposed by Tang (2012) is applicable in China. Our conclusion, methodology, and estimation results have important implications for various aspects of the Chinese economy including the regional, urban and real-estate economies.