In this paper we present a parallel implementation of T–Coffee — a widely used multiple sequence alignment package. Our software supports a majority of options provided by the sequential program, including the 3D–coffee mode, and uses a message passing paradigm to distribute computations and memory. The main stages of T–Coffee, that is library generation and progressive alignment, have been parallelized and can be executed on distributed memory machines. Using our parallel software we report alignments of data sets consisting of hundreds of protein sequences, which is far beyond the capability of the sequential T–Coffee program.