A fixed bed maleic anhydride reactor is simulated using an improved reactor model. A simplified Langmuir-Hinshelwood-type reaction kinetics is used, and the kinetic as well as the reactor parameters (for a full-scale pilot plant reactor and an industrial reactor) are tuned. The model is then used to solve a few optimization problems involving a single and multiple objectives from among maximum productivity, minimum operating cost, and minimum pollution. This is carried out using NSGA-II-aJG. A semiempirical procedure is suggested to reduce the scatter in the Pareto optimal solutions obtained for the three-objective optimization problems, and replace solutions which are associated with extreme sensitivity (this is different from obtaining a robust Pareto set, where all the optimal points have reduced sensitivity). Thus, our focus is twofold: modeling and multiobjective optimization of MA reactors, and improving the algorithm to deal with the extreme sensitivity of the three-objective problem (which may be of benefit in other problems as well). In addition, it is observed that a more recent algorithm, Alt-NSGA-II-aJG, biomimicking the altruism of honeybees, converges to the optimal solutions faster than does NSGA-II-aJG for a two-objective optimization problem, but it gives inferior solutions for a three-objective problem.