During an urban freight distribution, a number of dynamic incidents, such as traffic congestion, adverse weather conditions and mechanical failures may occur, leading to delays, time windows violations, higher costs, and inferior customer service. This paper presents the design of a real-time fleet management system that handles dynamically unforeseen events by detecting deviations from the initial distribution schedule and by recommending intelligent re-routing interventions in order to improve urban delivery performance. A system prototype was tested within real-life operations of two freight carriers in Greece demonstrating interesting results in a number of key performance indicators such as (time window violations, operational cost and total number of served clients).