In this work, we study, model, and propose two approaches to solve a raw milk transportation problem inspired by a real case of a milk company in Chile. The milk is produced by a set of farms scattered in a large rural area. The company must collect all the production daily using a truck fleet. We address the location of milk collection centers to reduce transportation costs. Each center has a limited capacity and a reduced truck fleet, composed of small trucks, to collect a substantial proportion of the produced milk. Once the milk is accumulated in the collection centers, a fleet of big trucks, traveling from a processing plant, collects the milk of each collection center and some large farms. We propose a mixed-integer linear programming model, a three-stage approach based on mathematical models, and an iterated local search approach to face this problem. We evaluate these approaches’ performance using a small case and several real-world examples, including a clustering approach to divide the instance into small sub-instances. The results obtained for the real-world instance show improvements of up to 10% percent when milk collection centers are allowed.