We analyze the minmax regret objective function for ellipsoidal uncertainty sets. The complexity of evaluating a solution and finding the best solution is investigated. We give complexity results for several combinatorial optimization problems. Two different cut generating strategies for an exact solution procedure are defined. Both are applied to unconstrained and shortest path problems in experiments.