This paper addresses a real-life logistic problem arising in the hospital complex of Tours (France). The two-level vehicle routing problem has time windows, a heterogeneous fleet, and multi-depot, multi-commodity and split deliveries. The first level concerns the routing problem for a fleet of vehicles serving several hospital units that delivers medicines, clean linen, meals, various supplies, patient files and picks up waste and dirty linen. The second level concerns the problem of routing employees between buildings within a large hospital unit. Both levels are interconnected. In addition, decisions about sizing and planning the teams of drivers and warehouse employees have to be made. Two metaheuristic algorithms are proposed to solve the entire problem: a genetic algorithm and a tabu search. The algorithms are tested on 100 instances, randomly generated on the basis of real-life instances.