For financial volatility modeling，most of the studies use returns as proxies of volatility，whereas very few are devoted to volatility methods based on range，which is a more efficient proxy． Taking the advantages of stochastic volatility method into consideration，this paper introduces the regime shifts of volatility levels into the range based stochastic volatility model to capture possible structural changes in volatility levels in financial markets． Afterwards，this paper describes the MCMC algorithm to estimate the model and demonstrates its efficiency through a simulation． In the empirical part，based on the range data of Shanghai Composite Index，Shenzhen Component Index and China Securities Index 300，the RMSSV model is estimated． Using the realized volatility as the benchmark and robust loss function as the criterion，the relative advantage of the RMSSV model in comparison with several popular models in GARCH and SV families is demonstrated．