Résumé : The increasing share of wind energy induces a strain on the electricity network. To unburden the transmission system operators from this strain, the dispensable wind energy can locally be stored in an energy carrier, e.g. ammonia (NH3). Existing work considers fixed operational parameters during design optimization to represent real-life conditions of the Power-to-NH3 system. However, uncertainties significantly affect real-life performances, which can lead to suboptimal plants. To provide a robust design—least sensitive to uncertainties—we considered the main operational uncertainties during design optimization and illustrated the contribution of each uncertainty on the systems NH3 production. This work presents the optimization under uncertainty of the Power-to-NH3 process and a global sensitivity analysis on the optimized designs. The results revealed a design trade-off, where a productive design produces 3.2 times more NH3 on average, but is 2.6 times less robust (higher standard deviation) than the robust design. A global sensitivity analysis on the most robust design showed that the temperature fluctuation of the NH3 reactor dominates the average NH3 production by 99.7%. The same sensitivity analysis on the most productive design showed that the wind speed measurement error and the temperature variation are both influencing the ammonia production by respectively 75.4% and 22.5%. Accordingly, an accurate anemometer and improving the temperature control over the NH3 reactor are the most effective actions to make the most productive design more robust. However, a robust plant can be obtained by decreasing the load size of the plant. It suffices to improve the temperature control over the NH3 reactor to make this design (adopted from the trade-off) less sensitive to the noise. Future investigations involve analyzing the dynamic operations of the robust Power-to-NH3 pathway and analyze the impact of uncertainties on its levelized cost.