This paper presents a validation experiment of a global CGE trade model widely used for analysis of trade liberalization. We focus on the ability of the model to reproduce price volatility in wheat markets. The literature on model validation is reviewed with an eye towards designing an appropriate methodology for validating large scale CGE models. The validation experiment results indicate that in its current form, the GTAP-AGR model is incapable of reproducing wheat market price volatility and that trade and poverty analysis taking these prices as inputs might be wrongly influenced. We conclude that model validation is tractable and an important component of analysis, as it points to areas to consider for refining analysis. In the specific case presented here, it seems that the Armington structure and trade elasticities are influential in whether the model over our under predicts price volatility for a specific region based on its net trade position for wheat.