The Balassa-Samuelson model, which explains real exchange rate movements in terms of sectoral productivities, rests on two components. First, for a class of technologies including Cobb-Douglas, the model implies that the relative price of nontraded goods in each country should reflect the relative productivity of labor in the traded and nontraded goods sectors. Second, the model assumes that purchasing power parity holds for traded goods in the long-run. We test each of these implications using data from a panel of OECD countries. Our results suggest that the first of these two fits the data quite well. In the long run, relative prices generally reflect relative labor productivities. The evidence on purchasing power parity in traded goods is considerably less favorable. When we look at US dollar exchange rates, PPP does not appear to hold for traded goods, even in the long run. On the other hand, when we look at DM exchange rates purchasing power parity appears to be a somewhat better characterization of traded goods prices.