The study explores a long memory conditional volatility model on international grain markets, demonstrating importance of modeling both temporal effects of volatility and long memory process. This study adopts six different volatility models, nested in an ARMA(p,q)- FIGARCH(P,D,Q), to capture dependence of grain cash price volatility and compares the performance of the six models. It also visits a related question about non-normal behaviors of grain prices and adopts the student-t density intended to account for fat-tailed properties of the data. We find suitability of the FIGARCH type models under the student-t distribution and competitiveness of the parsimonious FIGARCH(1,d,0) for modeling long memory volatility.