McCormick (1997) concluded that peripheral cues presented below a threshold of awareness could nevertheless attract attention because they facilitated target processing near the cue shortly after its presentation. Yet, whereas an exogenous shift of attention typically exhibits a biphasic pattern (initial facilitation followed by inhibition of return [IOR]), at late cue-target onset asynchronies, IOR was not observed by McCormick. In our study, targets requiring a detection response were preceded by masked and nonmasked, uninformative cues presented under two conditions: one in which the cue was ignored (no report) and one in which the cue was detected and localized following the response to the target (cue report). When participants were required to make cue judgments at the end of each trial, we replicated McCormick's pattern, finding facilitation (but not IOR) following both masked and nonmasked cues. When there was no requirement to judge the presence or location of the cues, IOR was present with and without masks, whereas facilitation was observed only when the cues were not masked. That the assessment of cue awareness increases attentional facilitation and prevents (or delays) the onset of IOR is attributed to attentional control settings put in place to perform the cue-awareness assessments in the cue-report condition.