This paper examines the relative importance of stochastic volatility and price jumps in option pricing by exploiting the differences in option-based and index return-based risk-neutral densities. As evinced by the persistent smirked volatility smile, this study extracts the risk-neutral densities by exploiting the information embedded in the implied volatilities that are smoothed by a kernel regression. On the other hand, a canonical valuation approach is adopted to identify the risk-neutral density from the observed index returns. Statistical tests and implied risk aversion are constructed to investigate the differences between two risk-neutral densities. The 30-day S&P 500 options under stochastic volatility are found efficiently priced. By using a longer horizon of underlying returns, however, we are able to partly reconcile the differences between the index and option-implied risk-neutral densities after adding a jump component to the index dynamics. Option investors are found more risk averse than stock traders except for the 30-day jump-diffusion with stochastic volatility. The finding may help illustrate the puzzle that implicit volatilities are greater than subsequent realized volatilities.