ABSTRACT This paper aims at exploring the asymmetric effect of the investor herding behaviour on the price bubble in the Chinese stock market. We first use the backward sup ADF (BSADF) statistic and the time-varying coefficient CCK (TVC-CCK) model to dynamically measure the stock market bubble and the investor herding. Then, we adopt the nonlinear autoregressive distributed lag (NARDL) model to analyse the asymmetric impacts of the positive and negative changes of the herding behaviour on the market bubble under different market conditions from the long and short run. The empirical results show that for the Chinese stock market, the investor herding has different influence mechanisms on the price bubble under different market states, and the effects of the positive and negative herding are asymmetric in the both long term and short term. In the bull market, the continuous rise of the herding plays a fuelling role in the formation of the bubble, and the weakening of the herding can significantly curb its expansion in the long run. However, in the bear market, the aggravation of the herding plays an ‘accelerator’ role in the collapse of the bubble, and the decline of the herding can’t significantly prevent this collapse.