The corrections to account for the changes in the cross-sectional area and for the restraint of the membrane during a triaxial test may vary considerably depending on the actual behavior of the soil specimen. This paper shows how observations and measures can lead to a choice of corrections which are in good agreement with the true behavior of the soil-membrane system both for bulging failures and for failures along a shear plane. The data obtained by compression tests on rubber and rigid dummies offer a valuable basis for quantifying the required area and membrane corrections. An example illustrates the merits of the proposed corrections and shows that different interpretations of the membrane correction may have an important implication on the effective stress cohesion intercept.