This paper revisits the theory and the empirics of the classic Balassa-Samuelson productivity model of cross-country differences in price levels. The classic model says that price levels are positively related to productivity in the traded sector and negatively related to productivity in the non-traded sector. I show that once endogenous specialization and costly intra-temporal trade are added to the classic model, this result is modified in two ways. First, the elasticity of relative prices with respect to traded sector productivity depends on the strength of terms-of-trade effects, and may even be negative if terms-of-trade effects are sufficiently strong. Second, the asymmetric trade patterns induced by costly trade mean that the relative price level between a pair of countries depends on productivity in all other countries that these two countries trade with. I simulate the model with trade costs, and show that both of these effects are potentially important. I then use the insight that traded-sector production is specialized to reassess the empirical evidence on productivity models of price level differences. I show that the evidence of other researchers is inconclusive. I estimate the relationship in a way that is consistent with the theoretical model. The model is not rejected, but the evidence is not strongly in favor either.