The welfare cost of Japanese rice policy is estimated in the context of a large importing country, treating domestically produced rice and imported rice as heterogeneous goods and where there is home‐good preference. Not accounting for this preference will cause the gains from liberalisation to be overestimated. The period that is analysed is 2004–2007, departing from that in previous studies, which do not cover this period of greater deregulation. Rather than use border trade flow data as is customary, we acknowledge the actions of a state trading enterprise and construct and use a unique data set which should better gauge import penetration in the Japanese market for rice. Econometric estimation fails to reject the hypothesis that the Ministry of Agriculture, Forestry and Fisheries prevents imports from affecting the price of domestically produced rice. In the absence of precisely estimated parameter values, simulations of liberalisation are conducted under a range of parameter values and the effects on social welfare calculated. The tariff equivalents of the government's support to rice producers are also estimated with values for the period in excess of 100 per cent.