An important stylized fact in the empirical Free Trade Agreement (FTA) literature is that member trade flows gradually increase over time following an FTA. Baier and Bergstrand (2007) suggest two explanations: tariff phase-out and delayed pass-through of tariffs into import prices. We examine these hypotheses using 1989–2016 U.S. import growth and product-level data on the tariff phase-out negotiated under NAFTA and the earlier Canada-U.S. FTA. We do not find evidence supporting either hypothesis. While products receiving tariff cuts do show delayed import growth relative to products with unchanged tariffs, the delay in import growth does not correspond to delays in the timing of tariff cuts. We also show that tariff cuts are fully and immediately passed through to U.S. importers as there are virtually no changes in the prices received by exporters either in the short run or the long run. Rather, we find evidence for an important role played by NAFTA tariff cuts reducing the impact of frictions that, in turn, allow for a spatial expansion of imports across the U.S.