ABSTRACTThis article provides a critical analysis of the models and methods which have been used to estimate the effects of economic integration on trade and suggests the most fruitful lines of further development. Both the problem of the examination of the effects of past events and problems in the prediction of future changes are tackled. In particular the drawbacks of the method of estimation of the effects of previous integration as a residual from models which explain what would have happened had the integration not taken place are shown. It is concluded that the most fruitful avenue of approach lies in the estimation of models which provide an economic explanation of trade flows and their changes, and can hence be used for both the explanation of the effects of previous integration and the prediction of future events.