The Electric Power Research Institute (EPRI) and GPU Nuclear Corporation have completed a demonstration project that provides justification for relaxing the high-pressure setpoints for the Oyster Creek Nuclear Generating Station. The project was undertaken because an undesirable overlap had been identified in the high-pressure setpoints when accounting for measurement uncertainties experienced during plant operation. The project employed a statistical combination of uncertainties (SCU) process to provide increased margin for measurement uncertainties. This approach was used because previous experience indicated that there was insufficient margin to justify the desired setpoints using conventional deterministic inputs to the safety analysis and plant performance analysis processes. Through the use of SCU methodology and other deterministic analyses, it is possible to provide comprehensive bases for the desired technical specification changes to the high-pressure setpoints. The SCU process is based on the EPRI setpoint analysis guidelines, and it requires the development of response surfaces to simulate RETRAN peak pressure calculations for the limiting transient events. The use of response surfaces adds an intermediate step to the SCU process, but reduces the number of RETRAN cases required to make appropriate statistical statements about the result probabilities. Basically, each response surface is an approximation of the RETRAN code for one particular event and one output variable of interest, which is valid over a limited region. The response surfaces can be sampled very inexpensively using simple Monte Carlo methods. The basic input to the development of a response surface is a set of results obtained from specific RETRAN cases. Each case includes a particular set of parameters consistent with an experimental design selected to ensure that all of the parameter dependencies are carefully considered and that the response surface fit has a reasonably small fitting error. The parameters selected for incorporation into the response surface are identified through a screening process that uses RETRAN analyses to establish the sensitivity of the event results to the parameter uncertainty. The parameter screening process, the selection of the experimental design, and the development of the response surfaces are described, and the analysis results are provided.