Pareto front optimization has been commonly used for balancing trade-offs between different estimated responses. Using maximum likelihood or least squares point estimates or the worst case confidence bound values of the response surface, it is straightforward to find preferred locations in the input factor space that simultaneously perform well for the various responses. A new approach is proposed that directly incorporates model parameter estimation uncertainty into the Pareto front optimization. This step-by-step approach provides more realistic information about variability in the estimated Pareto front and how it affects our decisions about the potential best input factor locations. The method is illustrated with a manufacturing example involving three responses and two input factors.