We initiate research on self-stabilization in highly dynamic identified message-passing systems where dynamics is modeled using time-varying graphs (TVGs). More precisely, we address the self-stabilizing leader election problem in three wide classes of TVGs: the class T C B (∆) of TVGs with temporal diameter bounded by ∆, the class T C Q (∆) of TVGs with temporal diameter quasi-bounded by ∆, and the class T C R of TVGs with recurrent connectivity only, where T C B (∆) ⊆ T C Q (∆) ⊆ T C R . We first study conditions under which our problem can be solved. We introduce the notion of size-ambiguity to show that the assumption on the knowledge of the number n of processes is central. Our results reveal that, despite the existence of unique process identifiers, any deterministic self-stabilizing leader election algorithm working in the class T C Q (∆) or T C R cannot be size-ambiguous, justifying why our solutions for those classes assume the exact knowledge of n. We then present three self-stabilizing leader election algorithms for Classes T C B (∆), T C Q (∆), and T C R , respectively. Our algorithm for T C B (∆) stabilizes in at most 3∆ rounds. In T C Q (∆) and T C R , stabilization time cannot be bounded, except for trivial specifications. However, we show that our solutions are speculative in the sense that their stabilization time in T C B (∆) is O(∆) rounds.