This paper deals with determining vehicle routes in product distribution from a distributor to a large number of retailers in a vast distribution area. To determine the vehicle routes in a vast distribution area and a large number of retailers that have to be served causes the VRP model to become complicated. Therefore, this paper applies the cluster first-route second concept to design a distribution system in two stages. The first stage is to cluster the distribution area based on the distance between retailers. The second stage is to determine a vehicle route for each cluster using the multi-objective vehicle routing problem with time window and balanced driver workload (VRPTWBW) approach. The proposed VRPTWBW model has three objectives, i.e., (i) to minimize the number of assigned vehicles, (ii) to minimize the total delivery time, and (iii) to balance the driver workload. This second stage starts once the clusters of the distribution area are set. Contrasted with the current practice in the company, this proposed model offers the significant impacts: (i) the reduction of workload gap among driver about 29% (from 2.59 hours to 1.83 hours); (ii) the increased of vehicle utility by 44% (from 30% to 43%), and (iii) the increase savings of two-week course fees of IDR. 11 million (about 23.84%).