Although some papers argued that multi-sensor fusion could improve performances and robustness of fingerprint verification systems, no previous work explicitly dealt with such topic, and no experimental evidence has been reported. In this paper, we show by experiments that a significant performance improvement can be obtained by decision-level fusion of two well-known fingerprint capture devices. As, to the best of our knowledge, this is the first work on multi-sensor fingerprint fusion, we believe that it can contribute to stimulate further researches on this promising topic.