This study examines whether term structure can be used as a predictor of commodity price direction. It uses daily prices for the S&P GSCI and each of the 24 underlying commodities from January 2007 through December 2013. During each day of the monthly roll period, one-month returns for each commodity are calculated and compared with the corresponding term structure. In nearly all cases, the relationship between returns and term structure is no different from that of a random process. Mean returns for each commodity under backwardation-only and contango-only are examined and in most cases are not significantly different from zero. A detailed examination of unleaded gasoline finds that returns on a long position are little affected by term structure, but heavily affected by the price trend in the underlying market.