Based on Chinese monthly data from January 1996 to November 2013,this paper first discuss the construction methods of nonlinear financial conditions index (NFCI). Then we measure China NFCI using Markov switching impulse response method of Markov switching vector autoregression (MS-VAR) model and do empirical analysis on the ability FCI to predict prospective inflation. The results show in the analysis of the effect that financial variables have on inflation and how they can predict the prospective inflation, it is better to use the NFCI which can reflect many financial variables such as money supply, interest rate, exchange rate and stock prices than linear FCI. NFCI is the leading indicator of inflation in China which contains much information about prospective inflation. We can use NFCI to predict the trend of inflation within the next six months in our country effectively.