In the option pricing literature, many researchers have focused on information contained in stochastic volatility, such as the CBOE VIX index and its risk premium. However, there are relatively few studies that focus on stochastic skewness and its risk premium. In this paper, we document that stochastic skewness is a strong factor for cross-sectional index put option returns. As a theoretical motivation, we present a general option pricing framework. This framework summarizes the common structure shared by recent advanced option pricing models, and it highlights the channel through which skewness affects option prices. Guided by such framework, using Fama-Macbeth two-pass regression and a panel of S&P 500 index put option returns, we find that skewness has statistically significant positive risk premium. Compared with volatility, skewness shows better pricing performance for cross-sectional put option returns. In particular, all abnormal returns for put options become insignificant after the skewness factor is incorporated into the regression. Moreover, loadings on option returns for volatility and skewness factors show different patterns. The volatility factor loads on put option returns with almost the same magnitude, while the skewness factor loads heavily on out-of-money put options.