This paper proposes an original method for accurately measuring the stator leakage reactance of synchronous machines. The approach relies on an experimental equivalence of resulting fluxes between the unsaturated open-circuited and permanent short-circuited tests. This requires the presence of measurement turns in the air gap of the machine. A complete comparison with existing methods (Potier, El-Serafi & Wu, rotor removed test and geometric calculation) is moreover proposed. A critical analysis is performed, and pros and cons of each method are highlighted.