RFID is an automatic identification technology that enables tracking of people and objects. We propose a scalable and robust 3-D localization method of RFID tags, derived in two schemes, that uses the diversity of pervasive environments to improve its accuracy. RFID readers are placed on the floor and ceiling of a room. In the first scheme, static and mobile tags are localized using only connectivity information obtained through this structure. The second scheme involves in addition mobile RFID readers. They are used here to refine the localization process by adding more inclusive constraints to the calculation. We study the impact of the number of static and mobile readers on the accuracy. We then show that the first scheme constitutes an upper bound for the second one. Furthermore, we observe that the second scheme does not require a lot of mobile readers and that it considerably improves accuracy.