The present research measured participants' event-related brain activity while they performed a Stroop-priming task that induced the implementation of expectancy-based strategic processes. Participants identified a colored (red vs. green) target patch preceded by a prime word (GREEN or RED), with incongruent prime-target pairings being more frequent (75 %) than congruent pairs (25 %). The prime-target stimulus onset asynchrony (SOA) was manipulated at two levels: 300 vs. 700 ms. Participants also performed a change localization task to assess their working memory capacity (WMC). At the 300 ms SOA, all participants presented a Stroop-priming congruency effect (slower responses on incongruent than on congruent trials) and an increased N2 amplitude in incongruent trials, irrespective of their WMC. At the 700-ms SOA, the lower-WMC group showed again a larger negative-going waveform to incongruent targets, whereas the higher-WMC group exhibited a reversed Stroop-priming congruency effect (faster responses to incongruent targets) and the N2 component was absent.