A virtual power plant (VPP) is a system that integrates several types of power sources, so as to give a reliable and friendly overall power supply. The sources are often a cluster of distributed generation systems with intermittent renewable energies. Uncertainties are the important issues in researches and applications of VPP. In this paper, renewable power, market price and load demand are classified as major factors of uncertainties, and a comprehensive review of these three factors are given. Based on the classification, the detailed mathematical descriptions are summarized. And then, optimization objectives and constraints, which are adopted to improve the running performance of VPP with uncertainties, are summed up systematically. Solution approaches and tools for the optimization are also presented. At last, demonstration projects are introduced to show how uncertainties are handled in practice. This review paper can provide a rational assistance for researchers who focus on VPP.