This paper investigates convergence patterns among China's provinces using GDP data for the period 1952-97. We use the Markov chains proposed by Quah (1993) and the generalized entropy decomposition approach proposed by Shorrocks (1980, 1984) to analyse the convergence behaviour. Both sets of results show similar evidence of convergence within the predefined geo-economic sub-regions, but no evidence of convergence between the sub-regions. The finding has important policy implications on regional economic development in China.