We are reporting in this study the detection of 1888 corrosion defects using a the Magnetic Flux Leakage (MFL) intelligent pig tool, over 70 km of a pipeline located in Algeria. This large amount of defects has been statistically analyzed. The relative defect depth a/t exhibited a large scatter and no correlation was found between corrosion defect depth and length. For the necessity of repairing defect, two tools are available: the first is based on limit analysis and called the estimated repair factor (ERF) while the second is based on failure assessment diagram. The adopted tool in the current study was the domain failure assessment diagram (DFAD). Analysis made with FAD concerns 66.8% of corrosion defects, with a limit analysis (LA) of 32.5%. After categorizing the corrosion defect according to the used analysis tool, the safety factor or probability of failure of each assessment point was determined and compared to the repairing criteria. It appears that the ERF criterion is more conservative in our case than the probabilistic criterion as a probability of failure of 10−4 or a non-dimensional crack driving force equal to mean minus 3 standard deviations.