Chinese mutual fund investors tend to chase funds with better historical performance and higher idiosyncratic volatility, leading to huge losses when fund performance reverses in the future. By incorporating the uncertainty of the fund returngenerating process into the investor learning model, this paper theoretically shows that investors with insufficient financial literacy are subject to the optimism bias while assessing the fund’s skills. This behavior results in an asymmetric impact of a fund’s idiosyncratic volatility on fund flowperformance sensitivity: When a fund performs well (poorly), the higher the fund’s idiosyncratic volatility, the higher (lower) the fund flowperformance sensitivity. The results show that a fund’s idiosyncratic volatility can explain the fund flowperformance convexity puzzle: The higher the fund’s idiosyncratic volatility, the more significant the fund flowperformance convexity relationship. Furthermore, by decomposing a fund’s idiosyncratic volatility into a persistent component (longrun idiosyncratic volatility) and a shortterm volatile component (shortrun idiosyncratic volatility), the paper finds that only longrun idiosyncratic volatility has a significant impact on the fund flowperformance convexity, while shortrun idiosyncratic volatility does not. Longrun idiosyncratic volatility does not help to improve the predictive power of historical fund performance on future performance. It significantly impacts individual investors’ fund flows, suggesting that this effect is due to investors’ behavioral biases. The research in this paper has important policy implications: The regulation of fund companies’ marketing should be strengthened to prevent funds from going viral. More importantly, investor education should be vigorously strengthened to enhance investors’sophistication and help them circumvent the optimism bias when learning the fund’s skills for investors with limited financial literacy.