The technological development of autonomous vehicles (AVs) challenges their economic relevancy. Are AVs worth it? We adapted the cost–benefit analysis framework to the case of AV services. The appraisal methodology was applied to the introduction of an on-demand AV service in Berlin, simulated through MATSim, an agent-based model. The introduction of AVs in a dense urban environment would generate more pressure on the road network. This would cause private car users to experience longer travel times, resulting in a consumer surplus loss that would not offset the new AV users’ benefits. In one configuration, a stop-based routing service appeared to generate fewer externalities (e.g., congestion, emissions, accidents) and would be cheaper to operate than door-to-door routing. Another configuration assumed that private cars were banned. In that case, the consumer surplus significantly decreased, AV services were not able to absorb the entire demand, and the shares of other modes increased. However, the externalities were significantly reduced. Finally, in an urban environment where on-demand AVs were added to the available modes, the social impact was negative for all economic agents, with heterogeneity apparent on the impacts of the different externalities. This would benefit some travelers but would also involve additional congestion. In that case, a stop-based routing seemed preferable (but still negative) to a door-to-door one. Further works discussing the introduction of AVs into an urban environment should focus on line-based options such as automated bus services.