When looking for two targets embedded in a rapid sequence of visual presentations (RSVP), subjects often accurately report the first target (T1) but not the second (T2), perhaps due to preoccupation with T1 consolidation. This is known as ‘Attentional Blink’ (AB). We ask if the AB is multimodal, but avoid using a dual task paradigm, which may confuse AB and task switching effects. We record auditory Event Related Potentials (ERPs) to unattended frequent standard and rare deviant tones, while participants performed only a two-target visual RSVP task. The dependent measure is the mismatch negativity (MMN), a signature of involuntary change detection, the difference between the ERP responses to auditory standards and deviants. Surprisingly, we find that when conditional identification of visual T2 fails (for a T1-to-T2 lag of 3 stimuli), the auditory MMN is significantly increased compared to trials (with the same lag) where both visual targets are correctly reported. Thus, when visual attention falters, auditory processing gains. This could be due either to release of sensory attentional resources by the “blinking� visual system, allowing more resources to be devoted to audition, or to a failure to inhibit the processing of irrelevant auditory information during the blink period.