The uncertainties in system and model parameters arising from the volatility of market, weather and operating conditions pose a major challenge to the optimal short-term operation of cascaded hydropower systems. The dynamic operating environment resulting from the fluctuating parameters greatly impacts the scheduling of power generation and generating unit commitment in such systems. This article focuses on the development and implementation a novel rolling horizon robust online scheduling framework that utilizes stochastic optimization within a model-based feedback scheme to tackle the uncertainties in electricity prices, electric power demands, water inflows and plant model parameters. The efficacy of this approach is demonstrated through application to a variety of case studies for different types of uncertainty. Case studies demonstrate significant improvements in system performance with the proposed strategy, in terms of system economics and constraint satisfaction, over schedules generated without feedback or use of a nominal online scheduling scheme. (c) 2020 Elsevier Ltd. All rights reserved.