The Multi-Compartment Vehicle Routing Problem involves clients with a demand for different products and vehicles with several compartments to co-transport these commodities. We present a local search procedure that explores well-known moves (2-opt, cross, exchange, relocate), and exploits the mechanisms of neighbour lists and marking to speed up the searches. We combine the procedure with the Guided Local Search meta-heuristic to improve solution quality. Extensive computational results are reported to uncover when co-distribution by vehicles with multiple compartments is better than separate distribution with un-partitioned trucks. Sensitivities in key problem parameters including, client density and location of the depot, vehicle capacity, client demand and number of commodities are investigated.