A switching adaptive robustified anti-windup compensator is proposed in order to improve the performance in the case of (possibly time-varying) large uncertainties, where previous purely robust solutions would be too conservative. Given a set of candidate controllers, the switching algorithm tries to determine and switch in the loop the most aggressive controller for which a small gain stability condition is satisfied according to the measured data (i.e. for the uncertainty currently affecting the plant), thus reducing the conservatism associated to previous robust solutions. An example shows the enhanced performance of the new algorithm. © 2010 IFAC.