The production and delivery of rapidly perishable goods in distributed supply networks involves a number of tightly coupled decision and optimization problems regarding the just-in-time production scheduling and the routing of the delivery vehicles in order to satisfy strict customer specified time-windows. Besides dealing with the typical combinatorial complexity related to activity assignment and synchronization, effective methods must also provide robust schedules, coping with the stochastic perturbations (typically transportation delays) affecting the distribution process. In this paper, we propose a novel metaheuristic approach for robust scheduling. Our approach integrates mathematical programming, multi-objective evolutionary computation, and problem-specific constructive heuristics. The optimization algorithm returns a set of solutions with different cost and risk tradeoffs, allowing the analyst to adapt the planning depending on the attitude to risk. The effectiveness of the approach is demonstrated by a real-world case concerning the production and distribution of ready-mixed concrete.