Very little is known about how listeners incorporate “intonational” information in whispered speech during online language processing. We present data showing that listeners can incorporate information about boundary tones in whispered speech rapidly, but this process is complicated by additional structural bi- ases as well as by the fact that speakers do not produce cues to boundary tones consistently in whisper. Listeners, however, are able to adapt to these differences in order to correctly identify different boundary tones in whisper.