Although theoretical methods were proposed for tolerating multiple disk failures, their complexities are too high to be practically applicable. In this paper, we proposed a practical parity scheme for tolerating simultaneous triple disk failures in RAID architectures. We first formalized the problems with matrix operations. Our scheme is practical in the sense that it employs three redundant disks for tolerating triple disk failures. Furthermore, it requires simple arithmetic computations only, and it can easily be implemented on current RAID controllers.