The prediction of the stabilization regime of partially premixed H2/air flames above an injector is a main subject of interest for the development of gas turbines powered by hydrogen. The way the flame is stabilized has an impact on NOx emissions, thermal stress on the injector, and combustion stability. A model based on the triple flame speed is revisited and improved to predict flame stabilization using information gathered at cold flow conditions. The model is called TFUP for Triple Flame Upstream Propagation. According to this model, the flame can anchor to the injector only if a zone with a flammable mixture and a sufficiently low local flow velocity exists continuously from the lifted flame to the injector lips. The TFUP model is applied to the case of a H2/air coaxial, dual swirl injector in which both the hydrogen and air streams are swirled. Fuel is injected through the central channel and oxidizer through the external channel. Particle image velocimetry and one dimensional Raman scattering measurements made at strategic locations are used to predict the edge flame speed for which a lifted flame should re-anchor to the injector. These predictions made in isothermal conditions are compared to observations of the flame stabilization regime. The central flow of hydrogen can be mixed with methane and helium and the external flow of air with nitrogen in order to prescribe different theoretical values for the triple flame speed. Predictions agree well with the observations made for all swirl levels conferred to the central fuel stream, fuel injection velocities, hydrogen contents in the fuel mixture injected in the central channel, and nitrogen contents in the oxidizer annular channel tested in the study. Validity limits are also discussed. The methodology presented in this study provides a simple framework to predict flame stabilization on coaxial injectors that can optionally be equipped with swirl vanes.