The aim of this study is to examine the link between working memory capacity and the ability to exert cognitive control. Here, participants with either high or low working memory capacity (WMC) performed a semantic negative priming (NP) task as a measure of cognitive control. They were required to ignore a single prime word followed by a pattern mask appearing immediately or after a delay. The prime could be semantically related or unrelated to an upcoming target word where a forced-choice categorization was required. Each type of mask (immediate vs. delayed) appeared randomly from trial to trial. Results demonstrated that, when the ignored prime was immediately followed by the mask, neither of the groups (high or low WMC) showed reliable NP. In clear contrast, when the mask onset was delayed responses latencies were reliably slower for semantically related trials than for unrelated trials (semantic NP), but only for the high WMC group. The present results clearly demonstrate that semantic NP from single ignored primes depends on both the masking pattern that follows the prime (immediate vs. delayed mask), and on working memory capacity.