We study complexity of the model-checking problems for LTL with registers (also known as freeze LTL and written LTL ↓ ) and for first-order logic with data equality tests (written FO(∼, <, +1)) over one-counter automata. We consider several classes of one-counter automata (mainly deterministic vs. nondeterministic) and several logical fragments (restriction on the number of registers or variables and on the use of propositional variables for control states). The logics have the ability to store a counter value and to test it later against the current counter value. We show that model checking LTL ↓ and FO(∼, <, +1) over deterministic one-counter automata is PSpace-complete with infinite and finite accepting runs. By constrast, we prove that model checking LTL ↓ in which the until operator U is restricted to the eventually F over nondeterministic one-counter automata is Σ 1 1 -complete [resp. Σ 0 1 -complete] in the infinitary [resp. finitary] case even if only one register is used and with no propositional variable. As a corollary of our proof, this also holds for FO(∼, <, +1) restricted to two variables (written FO 2 (∼, <, +1)). This makes a difference with the facts that several verification problems for one-counter automata are known to be decidable with relatively low complexity, and that finitary satisfiability for LTL ↓ and FO 2 (∼, <, +1) are decidable. Our results pave the way for model-checking memoryful (linear-time) logics over other classes of operational models, such as reversal-bounded counter machines. automata, see e.g. [16,17,18,19]. However, the storing mechanism has a long tradition (apart from its ubiquity in programming languages) since it appeared for instance in real-time logics [20] (the data are time values) and in so-called hybrid logics (the data are node addresses), see an early undecidability result with reference pointers in [21]. Meaningful restrictions for hybrid logics can also lead to decidable fragments, see e.g. [22].Our motivations. In this paper, our main motivation is to analyze the effects of adding a binding mechanism with registers to specify runs of operational models such as pushdown systems and counter automata. The registers are simple means to compare data values at different points of the execution. Indeed, runs can be naturally viewed as data words: for example, the finite alphabet is the set of control states and the infinite alphabet is the set of data values (natural numbers, stacks, etc.). To do so, we enrich an ubiquitous logical formalism for model-checking techniques, namely linear-time temporal logic LTL, with registers. Even though this was the initial motivation to introduce LTL with registers in [12], most decision problems considered in [12,13,8] are essentially oriented towards satisfiability. In this paper, we focus on the following type of model-checking problem: given a set of runs generated by an operational model, more precisely by a one-counter automaton, and a formula from LTL with registers, is there a run satisfying the given formula? In our ...