Reaction times (RT) during the Sternberg memory paradigm generally increase with memory set size, but do not differ for positive and negative probe stimuli. Sternberg proposed that this indicated that short-term memory (STM) scanning is both exhaustive and serial. However, this notion has received much criticism, primarily because RT must also reflect response selection factors. Magnetoencephalographic (MEG) recordings of auditory alpha-suppression have previously demonstrated that suppression duration is correlated with set size, potentially providing a physiological index of memory scanning time related specifically to sensory cortices. The current study expands earlier research into this metric by separately analyzing positive and negative probes. Thirteen normal adults participated in an auditory Sternberg paradigm. Pure tones were presented in memory set/probe combinations where the probe had a 50 percent chance of being within the memory set, and RT and accuracy were measured. Magnetic alpha-band activity (8-12 Hz) was quantified for pre- and post-stimulus regions. Although RT did not differ for positive and negative probes, alpha-suppression duration was greater for negative probes than positive ones, potentially indicating that scanning time was slightly faster in the positive condition. This may indicate that STM scanning is serial, but self-terminates when matching occurs.