The use of coupled 3-D neutron kinetics/thermal-hydraulic system codes is becoming increasingly important as a means to accurately simulate plant transients and to analyze complex accident sequences which are characterized by a significant change in the power distribution of the reactor core and by interaction between the core and the plant. In the development of these large computer codes the verification and validation procedure is vital to establish how well the codes perform. Part of the code verification includes the comparison of results with those of other codes. This can most easily be done through the calculation of benchmark problems, but the accuracy of the results can only be determined through comparisons with experimental data. Since only a few adequate experimental data exist to validate coupled codes an inadvertent event at the NPP Goesgen - a load rejection to auxiliary power supply - was used for the validation of RELAP5/PANBOX. The comparison of calculated and measured data shows the ability of RELAP5/PANBOX to analyze this type of transient and validates in-core and ex-core detector models and certain aspects of the coupled core and thermal-hydraulic models. Similar investigations are being performed with TRAC-PF1/NEM. Code-to-code comparisons with RELAP5/PANBOX will help to estimate and further reduce modeling uncertainties.