Negated words take longer to recognize than non-negated words following sentences with negation, suggesting that negated concepts are less active. The present experiments tested the possibility that this reduced activation would not persist beyond immediate testing. Experiment 1 used a probe task and materials similar to those used in previous research but manipulated the timing of the probe. The negation effect was present at 0 ms, replicating previous studies, but not at 500 ms or 1,000 ms, suggesting that unlicensed negated concepts are initially reduced in activation but then reactivated by presuppositional processing. Experiment 2 produced similar results when activation was measured during ongoing comprehension and reading time was controlled, and Experiment 3 demonstrated that these effects occur when negation is unlicensed but not when it is licensed. These findings are consistent with the hypothesis that the reduced activation seen from unlicensed negation is short-lived.