The precision of measurements of the strong coupling constant using event-shape variables in e +e- annihilations is limited by theoretical systematic uncertainties. The uncertainties are related to missing higher orders in the perturbative predictions for the event-shape distributions. A new method is presented for the assessment of theoretical uncertainties in α s . This method evaluates the systematic uncertainty of the parameter α s from the uncertainty of the prediction for the distributions from which it is extracted. The perturbative uncertainties are calculated on a purely theoretical basis, without accessing measured distributions. The method is therefore especially suited for an unbiased combination of results from different observables or experiments. It is universal and can be applied to other processes like jet production in deep-inelastic ep scattering or in hadron collisions.