Efficient message dissemination is of utmost importance to propel the development of useful services and applications in Vehicular ad hoc Networks (VANETs).In this paper, we propose a novel adaptive system that allows each vehicle to automatically adopt the optimal dissemination scheme in order to fit the warning message delivery policy to each specific situation. Our mechanism uses as input parameters the vehicular density and the topological characteristics of the environment where the vehicles are located, in order to decide which dissemination scheme to use. We compare our proposal with respect to two static dissemination schemes (eMDR and NJL), and three adaptive dissemination systems (UV-CAST, FDPD, and DV-CAST). Simulation results demonstrate that our approach significantly improves upon these solutions, being able to support more efficient warning message dissemination in all situations ranging from low densities with complex maps, to high densities in simple scenarios. In particular, RTAD improves existing approaches in terms of percentage of vehicles informed, while significantly reducing the number of messages sent, thus mitigating broadcast storms.