This paper introduces the indices of herding behavior and market volatility based on multifractal spectrum． Employing the nonlinear Granger causality test proposed by Diks and Panchenko and the cross-correlation coefficient proposed by Zebende，we investigate the dynamic linkages between herding behavior and market volatility． The empirical results indicate that the relationship between herding behavior and market volatility in the Shenzhen stock market is nonlinear，rather than the simple linear linkage． The correlations and causality relationships between herding behavior and market volatility are largely different between before and after the recent financial crisis． Specifically，before the reform of the shareholders in 2005，herding behavior and market volatility could not affect each other． During the period of 2005 － 2007，herding behavior had positive effects on market volatility． However，after 2008，herding behavior had negative effects on market volatility，meanwhile market volatility also had negative effects on herding behavior． This new finding indicates that herding behavior does not always result in“positive feedback”effects as claimed in conventional studies and “negative feedback”effects are present sometimes．