A parallel Gray code optimization (PGCO) algorithm is proposed in this paper. The Gray code optimization (GCO) shares some similarities with genetic algorithm (GA) and evolutionary programming (EP). It uses a binary representation, but the only operator is the mutation of a number of bits. The evolving strategy utilizes the adjacency property of the Gray code. By controlling how many bits to flip, it keeps a balance between global search and local search. Another property of the GCO is that the population size is not fixed. It grows linearly with the dimension of the problem, which help to alleviate the curse of the dimensionality. In order to avoid the slow convergence of high dimensional problems, a parallel Gray code algorithm using message passing interface (MPI) was implemented. Its scalability in a Beowulf Windows Cluster was investigated.