Visual attention has been conceived of as a dynamic system that configures and maintains visual input filters for specific features, objects and/or locations depending on goals and the current task demands. Temporarily losing control of this input system has been proposed as a cause of the performance impairment known as the attentional blink (AB). According to this proposal (Di Lollo, Kawahara, Ghorashi, & Enns, 2005), when the visual system is faced with identifying two masked targets presented in a rapid sequence, accurate selection and processing of the first target (T1) induces a momentary loss of control over the maintenance of input filters. This loss of control results in the impaired selection and processing of the second target (T2). The purpose of the present work was to test whether cued reconfiguration of attentional sets can be used to reduce the AB. In these dual-task experiments, T1 was not only one of the critical targets, but it also served as a cue that was either an accurate predictor of the category of T2 or it was not predictive of the category of T2. The results indicated that the severity of the AB was reduced when T1 predicted the category of T2 relative to when T1 did not accurately predict T2. These results suggest that the dynamic process of establishing and maintaining attentional sets can be rapidly reconfigured to reduce the performance decrements typically observed during the attentional blink.