While information technology has led to increased market integration, in many markets there are still significant heterogeneity associated with factors such as product characteristics, market location, market lots, transaction mode and buyer and seller characteristics. Using a data set of 40 million observations of up-streams fish sales in Atlantic France, we use the rich attribute information to investigate the extent to which the price differences in a highly heterogenous market can be explained, with a particular focus on the importance of transaction mode. While 78% of the quantity is sold at auctions, the remainder are over-the-counter transactions and this share is increasing. Estimation of hedonic price regressions controlling for fish attributes, local market, buyer and seller heterogeneity show that there is a price differential of 1.7% between the two transaction modes. This result remains robust when taking into account the endogeneity of the sale method through an exact matching of auction and over-the-counter transactions.