Previous empirical studies find a negative and significant relation between risk measures and expected future stock returns. Using four risk measures, we document that the negative risk-return relation is more pronounced among firms that receive high levels of attention from investors, while a standard positive risk-return relation holds among stocks to which investors pay little attention. Regardless of our proxy for risk, we find that the magnitude and statistical significance of the risk-related puzzle monotonically decreases as we move from high to low levels of investor attention. These findings suggest that investor attention may play a central role in risk-related anomalies. We also find that the risk-related puzzle is prominent in stocks to which individual investors pay more attention but almost non-existent in stocks to which institutional investors pay attention.