The paper describes a framework for constrained flight control based on the use of recoverable sets that are chained together to guide constrained admissible transitions between trim points. The recoverable set is the set of all states for which there exists a control sequence such that the subsequent response is guaranteed to satisfy the imposed constraints. The constraints can reflect actuator range/rate limits, safety limits, as well as ranges of validity of the aircraft model. Because in aircraft loss-of-control situations, fast onboard computations are necessary, the approach to computing recoverable sets in this paper exploits linear discrete-time models (which can be generated via onboard system identification and reflect effects of failures and degradations) and recovery sequences generated, either by a stable linear finite-dimensional subsystem or through a reset of the dynamic controller states or of its set points. With this approach, only subsets of the full recoverable set can be computed, however, these recoverable subsets possess certain desirable control invariance properties, and their computations are simple; moreover, the onboard generation of the recovery sequence reduces to a low-dimensional quadratic programming problem. The applications of this approach to longitudinal and lateral linearized and nonlinear aircraft flight models are reported.