Previous studies have demonstrated that observers may ignore highly salient feature singletons during a conjunction search task through focusing the attentional window (Belopolsky, Zwaan, Theeuwes, & Kramer, 2007), or by the suppression of bottom-up information (Treisman & Sato, 1990). In the current study, observers’ eye movements were monitored while performing a search task in which a feature singleton was present and corresponded with the target at a chance level. With practice, observers were less likely to make an initial saccade toward the singleton item, but initial saccades directed at the target were likely throughout. Results demonstrate that, in an effort to ignore the singleton, observers were more likely to suppress bottom-up information than adjust the size of the attentional window.