The sharp dose gradients which are possible in intensity modulated proton therapy (IMPT) not only offer the possibility of generating excellent target coverage while sparing neighbouring organs at risk, but can also lead to treatment plans which are very sensitive to uncertainties in treatment variables such as the range of individual Bragg peaks. We developed a method to account for uncertainties of treatment variables in the optimization based on a worst case dose distribution. The worst case dose distribution is calculated using several possible realizations of the uncertainties. This information is used by the objective function of the inverse treatment planning system to generate treatment plans which are acceptable under all considered realizations of the uncertainties. The worst case optimization method was implemented in our in-house treatment planning software KonRad in order to demonstrate the usefulness of this approach for clinical cases. In this paper, we investigated range uncertainties, setup uncertainties and a combination of both uncertainties. Using our method the sensitivity of the resulting treatment plans to these uncertainties is considerably reduced.