Using daily data from 182 spatially separated U.S. cash markets for the years 2006-2011, I investigate price discovery for corn. With a large number of cash markets available, I take into account explicitly the issue of market selection, which has been neglected in previous work. I find that empirical results concerning price discovery based on corn cash and futures markets vary with selection of cash markets. The cointegration relationship between corn cash and futures prices only holds for 52 cash markets based on logarithmic prices. And the informational source roles of futures and cash prices are equal in the long run for 49 out of these 52 markets. In the short run, the unidirectional causality from cash to futures prices is most possible no matter whether the cash market is cointegrated with the futures market or not. While the vast majority of causal relationships are linear, the causality from futures to cash prices is more likely to be nonlinear, especially for cash markets cointegrated with the futures market. For quantitative measures of the relative contributions of the futures market and an associated cointegrated cash market to the price discovery process, information share model and common factor model draw the same conclusion qualitatively and find that the contribution of the futures market is more likely to be small than a cash market.