This paper examines the inherent conservatisms of alternative girth weld defect acceptance criteria from the 2007 API 1104 Appendix A, CSA Z662 Appendix K, and the proposed EPRG Tier 2 criteria. The API and CSA codes have the same empirical limit-load criteria, where it has previously been shown that the conservatism on the failure stress is ∼30 to 50 percent compared to pipe test data prior to applying any safety factors. In terms of flaw length, it was found that the API/CSA limit-load equation might allow a flaw of 5% of the pipe circumference, where the properly validated limit-load equation would allow a flaw of 75% of the circumference, i.e., a safety factor of 30 percent on load corresponded to a safety factor of 15 on flaw length for that example case. Similarly there are conservatisms in a proposed EPRG Tier 2 girth weld defect acceptance criterion. This proposed criterion was directly based on curved-wide-plate data to assure that toughness was sufficient to meet limit-load conditions for a curved-wide plate. However, the curved-wide plates are really an intermediate-scale test, and still require proper scaling to pipes of different diameters. The proposed Tier 2 EPRG allowable flaw length is 7T from a large database of curved-wide-plate tests with the a/t value of less than 0.5 (or a < 3mm), and the failure stress being equal to the yield strength of the base metal (also requires the weld metal overmatch the base metal strength, and the Charpy energy at the defect location have a minimum > 30 J and average > 40 J). However, the widths of those curved-wide-plate tests are typically a factor 5 to 12 times less than typical large-diameter pipes. The proper limit-load/fracture mechanics scaling solution would have the flaw length proportioned to the plate width, not the specimen thickness. Additionally, the proper limit-load solution for a pipe in bending gives a much larger tolerable flaw size at the yield stress loading than a plate or pipe under pure tension. Example calculations showed that the EPRG Tier 2 approach is conservative on the flaw lengths by approximately 9 for pure axial tension loading, and between 34 to 79 for a pipe under bending. Suggestions are presented for an improved procedure that accounts for proper limit-load solutions for pipe tests, effects of pipe diameter, effects of internal pressure, and also a much simpler approach to incorporate the material toughness than the 2007 API 1104 Appendix A Option 2 FAD-curve approach. The fracture analyses could evoke SENB, SENT testing, or have relatively simple Charpy test data to assess the transition temperatures to ensure ductile initiation will occur.