Deciding shipment size is important in freight transport: it depends on the logistical imperatives of shippers and the technical possibilities of carriers. Shipment size choice is also closely related to transportation mode; it is therefore important from a public policy perspective. The theory of optimal shipment size and mode choice is robust. There are many inventory-theoretical models of optimal shipment size, applied by shippers in operational contexts. However, none of them has been validated empirically over a large and heterogeneous population of shipments, so that they are virtually useless for modeling freight transportation demand. This is due in particular to the lack of adequate data. In this paper, the simple Economic Order Quantity (EOQ) model is assessed empirically on a national scale, over a heterogeneous population of shipments. Using the French ECHO database, which notably observes commodity flow rates between shippers and receivers, the EOQ shipment size specification is estimated. The validity of the EOQ model is confirmed. In addition, the dominant role of the commodity flow rate between the shipper and the receiver, and of commodity value density is revealed. The relationship between mode choice and shipment size is also highlighted.