A critical issue in the quality assurance of safety-related software is the ability of the software to produce “identical” results, independent of the host machine, operating system, or compiler version under which the software is installed. A study is performed using the VIPRE-01, FREY-01, and RETRAN-02 safety-related codes. Results from an IBM 3083 computer are compared with results from a CYBER 860 computer. All three of the computer programs examined are written in FORTRAN; the VIPRE code uses the FORTRAN 66 compiler, whereas the FREY and RETRAN codes use the FORTRAN 77 compiler. Various compiler options are studied to determine their effect on the output between machines. Since the Control Data Corporation and IBM machines inherently represent numerical data differently, methods of producing equivalent accuracy of data representation were an important focus of the study. The study identifies particular problems in the automatic double-precision option (A UTODBL) of the IBM VS FORTRAN 1.4.x series of compilers. The IBM VS FORTRAN version 2 compilers provide much more stable, reliable compilation for engineering software. Careful selection of compilers and compiler options can help guarantee identical results between different machines. To ensure reproducibility of results, the same compiler and compiler options should be used to install the program as were used in the development and testing of the program.