Temporal binding describes an illusory compression of time between voluntary actions and their effects. In two experiments, using stable, preexisting action-effect associations, we investigated whether motor identity prediction (prediction of the effect's identity) enhances temporal binding. Touch-typists performed keystrokes and were presented with congruent (corresponding letter) or incongruent (noncorresponding letter) effects after different intervals. Touch-typists estimated the interval between keystrokes and effects. In both experiments, interval estimates were shorter with congruent than with incongruent effects, indicating that motor identity prediction contributes to temporal binding when using stable, preexisting action-effect associations. The congruency effect disappeared over the time course of Experiment 1 (in which incongruent effects were three times more likely than congruent effects), whereas it remained stable in Experiment 2 (in which congruent and incongruent effects were equally likely). Thus, the impact of motor identity prediction on temporal binding is context-sensitive. Even with highly overlearned action-effect associations, participants seem very flexible in adapting their internal predictions about an effect's identity. They may cease to use previously acquired action-effect associations in contexts in which their predictions are less reliable, thereby diminishing the influence of motor identity prediction on temporal binding. (PsycInfo Database Record (c) 2025 APA, all rights reserved).