The role of adequate non-destructive inspections and following repairs of detected flaws is essential in order to continue the integrity of welded structures. It is, however, pointed out that catastrophic failures sometimes happen due to insufficient repairs having no regard for unfavorable effects of repair. In case of welded joints, when a detected flaw is repaired by renewed welding, the fracture toughness of a flaw which is not discovered and adjacent to the detected flaw becomes smaller because of so-called the Hot Straining Embrittlement. Therefore, if such effect mentioned above is not taken into account, the reliability of the welded joint must be overestimated. The purpose of the research is to investigate the effect of the Hot Straining Embrittlement giving rise to serious deterioration of fracture toughness and evaluate the optimum range of repair in order to prevent an unexpected failure with the aid of reliability analysis. This report, at first, describes deterministic or probabilistic factors which should be adopted in the reliability analysis, a model and formulations of the analysis, and then, presents the comparison between estimated reliabilities of various cases, i. e., no repair, repair with or without the Hot Straining Embrittlement (HSE). From the results of numerical examples, it is shown that, 1) the probability of failure of repair with HSE is extremely larger than that of repair without HSE, 2) when the detected flaw is small, the probability of failure of no repair becomes smaller than that of repair with HSE, and 3) the important results are obtained on the range of repair with HSE.