This paper addresses the problem of set-point tracking for a continuous-time process modeled by a multi-input multi-output (MIMO) linear system that may switch, in unknown and unpredictable fashion, among different modes taken from a finite set. The proposed methodology relies on an high level controller which, from time to time, can switch on in feedback with the process a set-point controller, from a family of candidate controllers, based on a real-time estimation of the current process mode. It is shown that, under certain conditions, global exponential stability can be achieved for any slow-on-the-average process mode switching sequence.