The anti-windup (AW) problem is formulated in discrete time using a configuration which effectively decouples the nominal linear and nonlinear parts of a closed loop system with constrained plant inputs. Conditions are divided which ensure an upper bound on the induced l2 norm of a certain mapping which is central to the anti-windup problem. Results are given for the full-order case, where a solution always exists, and for static and low-order cases, where a solution does not necessarily exist, but which is often more appealing from a practical point of view The anti-windup problem is also framed and solved for continous-time systems under sampled-data control. It is proved that the stability of the anti-windup compensator loop is equivalent to a purely discrete-time problem, while a hybrid induced norm is used for performance recovery. The performance problem is solved using linear sampled-data lifting techniques to transpose the problem into the purely discrete domain. The results of the paper are demonstrated on a flight control example.