Abstract Material ductility, or the failure strain, is generally characterized by simple tensile tests. However, tensile tests on sheet samples lead to higher strain gradients than those observed in axisymmetric samples, so the classical Bridgeman correction is invalid. This paper first describes how to correctly calculate local values of stress the triaxiality in tensile deformed sheet samples. Secondly, the ductility of sheet samples was also simulated numerically by a damage mechanics cell model based on the finite element method. Finally, an experimental study on two Al 6xxx sheet alloys shows that the classical, average ductility (ln( A r / A 0 ) ≅ 0.67 here) strongly underestimates the true maximum material ductility, i.e. the local maximum strain to failure ɛ f ≅ 0.91. The values of the average ductility predicted by the present work are also shown to agree well with the experimental values.