Although trade liberalisation has been linked econometrically and through casual empiricism to large income increases, attempts to quantify its impact in static simulation models have shown small estimated gains. This paper shows that when the endogenous dynamic effects of trade liberalisation are built into simulation models, the estimated gains are indeed very large. In the authors’ model, intermediate inputs affect aggregate output through a Dixit-Stiglitz-Ethier function so that additional varieties provide the engine of growth. But complementary regulatory, financial market and macroeconomic reforms are important to realise the largest gains. The authors apply the model to five developing countries. In this model with lump sum revenue replacement, reducing a tariff from 20 percent to 10 percent produces a real income increase (in terms of Hicksian equivalent variation over the infinite horizon) of 10.7 percent of the present value of consumption in their central model, where the economy is assumed to be unable to borrow on international financial markets. If macroeconomic and financial reforms are in place that would allow international borrowing, however, the same tariff cut is estimated to result in a 37 percent increase in Hicksian equivalent variation. On the other hand, if inefficient replacement taxes must be used in an economy without the capacity to borrow internationally, the gains would be reduced to 4.7 percent. Larger tariff cuts (typical of those in many developing countries over the past 30 years) produce larger estimated welfare gains at least proportionate to the size of the cut.