As part of the European policy goals aiming towards a sustainable energy supply, the share of renewable energy sources in the electricity system is increasing. The two technologies driving this development in Europe, i.e. wind and PV, are characterised by a variable energy output, facing a limited predictability. These unexpected output variations challenge the real-time balance between electricity demand and supply and require appropriate operating reserve capacity. In the absence of large-scale storage or price responsive demand, the balancing of the power system is the main bottleneck for a large-scale deployment of renewable electricity generation in the power system.This dissertation deals with the management of operating reserves for balancing the power system with high shares of wind power. Firstly, a numerical model is developed to generate realistic time series of wind power generation and predictions over a region. This data is used to determine the short-term variations and prediction errors impacting the system balance. These are assessed by means of a flexibility assessment tool quantifying the operational thermal flexibility which is available for covering real-time system imbalances. Results show that without additional reserve requirements, this flexibility is insufficient to ensure a reliable integration of wind power. Particularly the upward fast-response flexibility is found to be scarce.Secondly, a statistical methodology is used to size and allocate additional reserve capacity to maintain stable reliability standards while integrating wind power. A strategy which minimises total capacity while maximising the allocation of reserve capacity towards slow-response reserves is put forward as a cost-efficient strategy. This is verified by means of a model simulating the day-ahead scheduling of power plants to meet the demand for power and reserve capacity. It is found that reserve requirements, and in particular the fast-response upward reserves, impact electricity generation cost. This cost increase is minimised when deploying slow-response reserve strategies. Results indicate the importance of peak power plants and base load flexibility to obtain a cost-efficient procurement of thermal reserve capacity.Finally, two reserve strategies are presented based on probabilistic wind power forecasting tools. Information concerning the uncertainty of the wind power forecast is used to obtain time-varying reserve requirements and enables the active participation of wind power in reserves. It is shown that both strategies achieve substantial cost savings without impacting system security. In particular the dynamic reserve strategy is particularly useful for reducing upward reserve capacity. In contrast, wind power participation remains a last-resort measure when facing low downward flexibility and elevated reserve requirements.