The development of renewable energy represented by wind, photovoltaic and hydropower has increased the uncertainty of power systems. In order to ensure the flexible operation of power systems with a high proportion of renewable energy, it is necessary to establish a multi-scenario power system flexibility evaluation method. First, this study uses a modified k-means algorithm to cluster operating scenarios of renewable energy and load to obtain several typical scenarios. Then, flexibility evaluation indices are proposed from three perspectives, including supply and demand balance of the zone, power flow distribution of the zone and transmission capacity between zones. Next, to calculate the flexibility evaluation indices of each scenario—and according to the occurrence probability of each scenario—we multiplied the indices of each scenario by the scenario occurrence probability to obtain comprehensive evaluation indices of all scenarios. Based on the actual historical output data of renewable energy and load of a southern power system in China, a flexibility evaluation was performed on the modified IEEE 14 system and modified IEEE 39 system. The results show that the proposed clustering method and flexibility indices can effectively reflect the flexibility status of the power system.