Since material handling agents in a FMS are most efficient when moving in straight lines, a common layout of machines in a FMS is a single row layout. This allows a floor designer to model the problem of generating an optimal machine layout in a FMS as a single row facility layout problem (SRFLP). Due to the computational complexity involved in solving the SRFLP, researchers have developed several heuristics to solve large instances of the problem. In this paper, we present a tabu search implementation based on a 3-opt neighborhood search scheme. We also present a technique to speed up the exhaustive 3-opt neighborhood search process significantly. Our computational experiments show that speed up of the 3-opt search is effective, and our tabu search implementation is competitive. The results we present here are better than the currently known best layouts for several large sized benchmark SRFLP instances, and competitive for other benchmark instances.