This paper examines the causal structure among the daily corn futures and seven cash price series from Midwestern states from January 3, 2006, to March 24, 2011, through a rolling approach that takes into account window sizes of a half, one, one and a half, and two years. Except for some testing samples, all series are tied together through cointegration and adjust toward the long-run relationship(s). Considering different forecasting lengths, the out-of-sample Granger causality test for each window generally reveals that no series gains persistent forecastability from another. These results shed light on the evolving causal structure among the different series. Discussions of empirical findings at a more granular level also are presented.