In this paper, we employ the Hilbert-Huang Transform to investigate the multifractal character of Chinese stock market based on CSI 300 index. The measured Hilbert moment Lq(ω) shows a power-law behavior on the range 0.01 < ω < 0.1min-1, equivalent to a time scale range 10 < τ < 100 min. The measured scaling exponents ζ(q) is convex with q and deviates from the value q / 2, implying that the property of self-similarity is broken. Moreover, ζ(q) and the corresponding singularity spectrum D(h) can be described by a lognormal model with a Hurst number H = 0.50 and an intermittency parameter μ = 0.12. Our results suggest that the Chinese stock fluctuation might be captured well by a multifractal random walk model with a proper intermittency parameter.