The NRC Regulatory Guide 1.99 (R.G. 1.99) Revision 2 was based on 177 surveillance data points and the EPRI data base, where 76% of 177 data points and 60% of EPRI data base were from Westinghouse's data. Therefore, other vendors' radiation environment may not be properly characterized by R.G. 1.99's prediction. To minimize scatter from the influences of the irradiation temperature, neutron energy spectrum, displacement rate, and plant operation procedures on embrittlement models, improved embrittlement models based on group data that have similar radiation environments and reactor design and operation criteria are examined. A total of 653 shift data points from the current PR-EDB, including 397 Westinghouse data, 93 B&W data, 57 CE data, and 106 GE data, are used. A nonlinear-least-squares fitting FORTRAN program, incorporating a Monte Carlo procedure with 35% and 10% uncertainty assigned to the fluence and shift data, respectively, was written for this study. In order to have the same adjusted fluence value for the weld and plate material in the same capsule, the Monte Carlo least-squares fitting procedure has the ability to adjust the fluence values while running the weld and plate formula simultaneously. Six chemical components, namely, copper, nickel, phosphorous, sulfur, manganese, and molybdenum, were considered in the development of the new embrittlement models. The overall percentage of reduction of the 2-sigma margins per ▵RTNDT predicted by the new embrittlement models, compared to that of R.G. 1.99, for weld and base materials are 42% and 36%, respectively. Currently, the need for thermal annealing is seriously being considered for several A302B-type RPVs. From the macroscopic view point, even if base and weld materials were verified from mechanical tests to be fully recovered, the linking heat-affetcted zone (HAZ) material has not been properly characterized. Thus the final overall recovery will still be unknown. The great data scatter of the HAZ metals may be the result of the metallurgical heterogeneity that exists in the HAZ. The proposed data fitting procedure for the HAZ material is presented in the paper.