This study investigates whether Mandarin listeners integrate a prosody-covarying phonological variable, the Tone 3 sandhi (T3S), into auditory sentence disambiguation. The Mandarin T3S process changes the first of two consecutive low tones (T3) into a rising tone. It applies obligatorily within a foot and optionally across feet. When T3S is optional, it is more likely to apply to T3 syllables across smaller prosodic boundaries than larger ones; the smaller the boundary, the sharper the T3S pitch rise. Participants listened to twenty-seven structurally ambiguous sentences containing two consecutive T3 syllables. Posing different T3-intervening prosodic boundaries would result in different interpretations. The first T3 syllable was manipulated into three tone shapes (sharp-rising, shallow-rising, low) and two duration types (long, short). Participants identified from two written interpretations the one consistent with what they heard. The results show higher major-juncture interpretation rates when the first T3 is long than short, when T3S does not apply than when it applies, and when T3S has a shallower than sharper pitch slope. The tone effect further interacts with the possibility of T3 syllable foot formation of each sentence. We propose that listeners have a sophisticated knowledge of prosodic variables and use it efficiently in linguistically meaningful contexts.