Welds with equivalent behavior to base metal performance can be produced today, even when the post weld thermal processing involves normalizing after welding. Alternatively, welding can result in properties far below typical base metal performance. Meaningful prediction of the service capabilities of weldments often requires testing specimens of the right size and configuration at the right stress. Performance relative to the reference lines depends on the specimen size, test stress and prior thermal history. Large valid tests can be performed in reasonable times to disclose the relative behavior of welds and base metal. However, large test samples, as discussed in this paper, must be used to obtain meaningful results. Standard tests at design level stresses are certainly practical for some materials. They have disclosed differences in service capabilities of welds of a factor of four (4) or more. Precise life prediction for weldments is subject to the same uncertainties as described for base metal. However, if there can be confidence in the pertinence of the test results, then suitability for continuing operation may be determined. Unfortunately, a great deal of the data available today on weldments cannot be taken at face value and conclusions drawn may be misleading. There is evidence that precursors to impending weldment failure may not appear until very late in life and not at the weldment surface. This calls into question much of the NDE (Non-Destructive Examination) which has been applied without adequate metallurgical consideration.