Experimental evidence suggesting that intertemporal preferences are time inconsistent is widespread. In particular, the "short-run" is found to be more heavily discounted than the "long run." Hyperbolic discounting is used to describe such preferences. This paper employs asset market data to test for non-exponential discounting. Our approach relies on state prices recovered using S&P500 index options. We find statistical evidence for the existence of hyperbolic time preferences.