We find that trade and domestic market size are robust determinants of economic growth overthe 1960-1996 period when trade openness is measured as the US dollar value of imports andexports relative to GDP in PPP US$ ('real openness'). When trade openness is measured asthe US dollar value of imports and exports relative to GDP in exchange rate US$ ('nominalopenness') however, trade and the size of domestic markets are often non-robust determinantsof growth. We argue that real openness is the more appropriate measure of trade and that ourempirical results should be seen as evidence in favor of the extent-of-the-market hypothesis.