Three procedures are used to test Fama semistrong from efficiency of harvesttime price of Kansas City July wheat futures from 1947 through 1995. The three methods are (a) testing for jointly significant parameter estimates on nonfutures explanatory variables in econometric forecasting models, (b) testing the relative accuracy between model-based forecasts and using deferred futures prices as forecasts, and (c) testing for abnormal profits associated with simulated futures trading signaled by the forecasts. Kansas City July wheat futures are generally efficient. Furthermore, relative to the efficiency associated with forecasts constructed one to two months before harvest, the efficiency associated with the five- to six-month period before harvest has increased, especially since the early 1980s.