To meet the required flight simulator-based stall recovery training that will become mandatory for all aircarrier pilots in 2019, many flight simulators will need to be updated with accurate flight models at high anglesof-attack. Using a database of 69 symmetrical and quasi-steady recorded stalls performed with TU Delft's Cessna Citation II laboratory aircraft, this paper aims to show to what extent relevant stall characteristics can still be modeled using flight test data from such "natural" stall flight test data (i.e., without additional flight test inputs). For modeling the (low-frequency) changes to the aerodynamic characteristics, the well-known stall model structure based on Kirchoff's flow separation theory is used. This aerodynamic stall model is augmented with a high-frequency stall buffet model identified based on power spectral density analysis of the flight test linear acceleration data. The stall model identification results obtained from a proposed methodology in which the aerodynamic and buffet model parameters were deliberately jointly estimated shows that the dynamic parameters capturing the stall hysteresis effect can indeed be estimated using quasi-steady stall maneuvers. Aerodynamic terms related to the pitch rate, however, are difficult to estimate for such quasi-steady maneuvers, given the correlation between pitch rate and angle-of-attack. It was found that estimating stall transient effects (i.e., hysteresis time constants), which normally require highly dynamic flight test maneuvers, was found to be improved with with explicit use of the measured accelerations caused by the stall buffet in the identification methodology.