The stretching model of McKenzie is applied to the formation of passive continental margins, assuming local isostatic equilibrium. We present the quantitative implications of the model; we then discuss its fit to the IPOD data on the Armorican and Galicia continental margins of the northeast Atlantic. The amount of brittle stretching observed in the upper 8 km of the prestretched continental crust reaches a maximum value of about 3. This large amount of thinning is camparable to the thinning of the whole continental crust observed by seismic refraction measurements and required by the model for the whole lithosphere. This agreement suggests that the simple stretching model is a good first approximation to the actual physical process of formation of the margin. It is thus possible to compute simply the thermal evolution of the margin and to discuss its petrological consequences. It is also possible to obtain a quantitative reconstruction of the edge of the continent prior to breakup. Finally, the large slope of the base of the lithosphere during the formation of the margin results in a force similar but opposite to the 'ridge-push' force acting on accreting plate boundaries.