This study investigates vehicle routing where some of the node attributes of location demand and time of availability are unknown at the start of the work cycle. The situation is dynamic with an element of stochasticity. Coordination in dynamic vehicle routing is implemented by estimating unknown attributes and developing contingency plans to cope with differences between estimated and actual values. Such implementation can have a beneficial influence on the costs of vehicle routing by reducing the total travel time or by rendering unnecessary an increase in fleet size that would have catered for unknown demand.