Territorial emergency scenarios imply several interesting issues with regard to the development of distributed applications and systems. From complexity to dynamism and unpredictability, these issues must be faced with appropriate paradigms and tools, which must grant flexibility and adaptability. In this paper, we apply concepts and abstractions from the Service- Oriented Computing paradigm and from Software Agents to support the management of territorial emergencies. Rather than making a theoretical comparison, we analyze a proper case study and show where the advantages and drawbacks of the two paradigms are more evident. From these considerations, we conclude that agents better suit this scenario, but need still to learn from the emerging paradigm of SOC.