We study the performance under uncertainty of three renewable energy policy instruments: Tradable Renewable Quota (TRQ), Feed-In-Tariff (FIT), and Feed-In-Premium (FIP). We develop a stylized model of the electricity market, where renewables are characterized by a positive learning externality, which the regulator aims to internalize. Assuming shocks on the fossil-based electricity supply, renewables supply, or on total electricity demand, we analytically derive the conditions determining the instruments’ relative welfare ranking. Although we generally confirm the key role of the slopes of marginal benefits and costs associated with the policy, the specific ranking depends on which type of uncertainty is considered, and whether shocks are permanent or transitory. However, a high learning rate generally favours the FIT, while TRQ is mostly dominated by the other two instruments. These results are confirmed in a numerical application to the US electricity market, in which the FIP emerges as the most and TRQ as the least robust overall choice.