The recent literature on ego depletion and self-control is plagued with failed replications. It has been argued that publication bias might have inflated estimated effect sizes. Doubts go so far that the very existence of the ego-depletion effect has been questioned. We conducted two high-power tests of the ego-depletion effect, with samples in two different countries, including a habituation phase in the depleting task (“e-crossing task”). This addresses recent critiques on failed registered replications, which argued that habituation was essential to obtain depletion effects. We examined the effect on error rates, response times, interference scores for error rates and response times, and response-time variability in a subsequent Stroop task. There were no effects in general, except a significant difference in response-time variability, only in one of the samples, and in the opposite direction as predicted by the ego-depletion effect.