Since 2019, a key element of simulator-based training of airline pilots is stall training. A major and still largely open research question is which level of model fidelity is required for effective training. As part of an effort to answer this question, a model of the quasi-steady stall dynamics of a Cessna Citation II aircraft is identified from flight test data that was specifically collected for this research at an altitude of 5,500 m. To ensure any reductions in elevator and aileron effectiveness during stall were also explicitly measured, the test pilots used additional quasi-random flight test inputs. The considered stall model structure is based on Kirchoff's theory of flow separation. During identification, the nonlinear and linear parameters of the model are estimated in separate, recursively executed, steps. This separation enables the application of a semi-objective model structure selection method using multivariate orthogonal functions for the aerodynamic coefficients included in the model. This approach shows that stall-related effects should be included in the model equations for lift, drag, and pitch moment. Overall, it is found that the model parameters were consistently estimated from the flight test data and that the model accurately describes the aircraft's stall dynamics in the considered flight condition. The developed methodology is concluded to be well-suited for the direct identification of stall models from flight test data.Nomenclature Roman Matrix of regression variables * Specific force in * -direction [m/s 2 ] a Regression variable vector * Force/moment coefficient [-] Average chord length [m] * Angular moment of inertia around * -axis [kgm 2 ] Cost function value Mach number [-] Mass [kg] Amount of time samples Amount of terms in a model structure Roll rate [rad/s] Pitch rate [rad/s] 2