According to Muhlbauer [1], 25% of the accidents with pipelines are caused by corrosion. The aging of the pipelines lead to an increase in the frequency accidents associated with an increase in maintenance costs. To keep accident frequency low, maintenance costs increase mainly because the current methodologies to predict failure and analyse pipe integrity are generally, overly conservatives. The new inspection techniques coming forth and the computers progress, it have justified the investment in research and in the development of numeric methodologies to preserve pipeline integrity and assure the economical competitiveness of their operators. Nine tubular specimens of steel API 5L X60, with 2 ms nominal length, 323 mms and 9.53 mms thickness were used. Defects were fabricated using spark erosion to simulate corrosion. The defects had residual thickness of 3 mm (30% of the nominal thickness), 95.3 mms width (10 times the nominal thickness) and lengths between 250 and 525 mm. The pipe specimens were instrumented with high elongation strain gage rosettes for great plastic deformations and they were pressurized until bursting. For the Finite element models were built based on the geometries of the specimens. In this analysis it was used the solid element of eight nodes because they demanded a smaller processing time than the one of 20 nodes. Regarding the shell element, the solid element of 8 nodes represent better the defect`s geometry and gives the values of elastoplastic strain along the thickness. Initially a linear analysis was carried out. These analysis were following by non-linear analysis, where it had been used the real properties of the material in pipe. The values obtained in these analysis were confronted with the experimental values of the elastoplastic strain measured, with the objective of validating the numeric model, had being obtained good correlation. During the validation of the model the numeric rupture criterion was analyzed to be adopted. This analyses included an investigation on the influence of small thickness variations , the influence of the radius between walls of the defects, the optimum increment of pressure and the importance of the mechanical properties. This research does a complete analysis of the tubular specimens, getting to develop a methodology able to reproduce the experiment. All the factors that influence a non-linear analysis were considered and the conclusion was that the experimental rupture happens inside of a range of values whose limits depend on the rupture criterion adopted. It was verified that this range includes the pressure that causes equivalent Misess stress equal to the stress of materials in the first element and the numeric instability. This instability happens when all of the elements along the thickness, in any point of the defect, reaches the equivalent Misess stress equal to the true failure stress of the material. In the studied cases, this range had distant limits amongst themselves of the order of 0.4 MPa, below 2% of the rupture pressure.