Although the oxidation of water is efficiently catalysed by the oxygen-evolving complex in photosystem II (refs 1 and 2), it remains one of the main bottlenecks when aiming for synthetic chemical fuel production powered by sunlight or electricity. Consequently, the development of active and stable water oxidation catalysts is crucial, with heterogeneous systems(3,4) considered more suitable for practical use and their homogeneous counterparts more suitable for targeted, molecular-level design guided by mechanistic understanding(5-19). Research into the mechanism of water oxidation has resulted in a range of synthetic molecular catalysts, yet there remains much interest in systems that use abundant, inexpensive and environmentally benign metals such as iron (the most abundant transition metal in the Earth's crust and found in natural(20,21) and synthetic(22) oxidation catalysts). Water oxidation catalysts based on mononuclear iron complexes have been explored(9,12,16,18), but they often deactivate rapidly and exhibit relatively low activities. Here we report a pentanuclear iron complex that efficiently and robustly catalyses water oxidation with a turnover frequency of 1,900 per second, which is about three orders of magnitude larger than that of other iron-based catalysts. Electrochemical analysis confirms the redox flexibility of the system, characterized by six different oxidation states between Fe-5(II) and Fe-5(III); the Fe-5(III) state is active for oxidizing water. Quantum chemistry calculations indicate that the presence of adjacent active sites facilitates O-O bond formation with a reaction barrier of less than ten kilocalories per mole. Although the need for a high overpotential and the inability to operate in water-rich solutions limit the practicality of the present system, our findings clearly indicate that efficient water oxidation catalysts based on iron complexes can be created by ensuring that the system has redox flexibility and contains adjacent water-activation sites.