We consider the tool loading problem in flexible manufacturing systems (FMSs) in which each part visits only one machine for its entire process. In these FMSs, if a machine does not have on its tool magazine the tools needed for a part assigned to it, it can borrow tools from other machine(s) or a tool crib or use alternative tools. The problem is to assign parts and tools to machines to minimize the number of tool transfers among machines. A heuristic solution is developed considering similarity among parts. The heuristic is compared with an optimal solution (for small-sized problems) and with an existing heuristic (for medium to large-sized problems). Also presented are computational results that show the effects of two tool redundancy policies (using alternative tools and using multiple tools for each tool type) on system performance.