The integration of distributed energy resources into a virtual power plant (VPP) can realise the scale merit. This study presents a novel approach for determining the optimal offering strategy of a VPP participating in the day-ahead (DA), the spinning reserve (SR), and the real-time (RT) markets. To hedge against multi-stage uncertainties, the authors propose a robust mixed integer linear programming optimisation model that comprises four levels: (i) the optimal DA energy and reserve dispatch; (ii) the worst-case realisation of uncertainties involved in DA market energy prices, SR market capacity prices, stochastic power production, and called balancing power; (iii) the optimal RT energy re-dispatch; and (iv) the worst-case realisation of uncertain RT market energy prices. Moreover, a tractable solution method based on strong duality theory and the column-and-constraint generation algorithm to solve the proposed four-level formulation was developed. Finally, numerical results for a realistic case study demonstrate the efficiency and applicability of the proposed approach. The commercial benefits of this strategy are also evaluated.