This paper contributes to the small but growing body of literature which tries to explain why, despite the predictions of some theoretical studies, empirical support for the pollution haven hypothesis remains limited. We break from the previous literature, which tends to concentrate on US trade patterns, and focus on Japan. In common with Ederington et al.'s (2005) US study, we show that pollution haven effects are stronger and more discernible when trade occurs with developing countries, in industries with the greatest environmental costs and when the geographical immobility of an industry is accounted for. We also go one step further and show that our findings relate not only to environmental regulations but also to industrial regulations more generally.