Many tone3 words in Mandarin undergo "third tone sandhi" - a phonological rule that changes the first tone3 word in a tone3+tone3 sequence to a tone2 word. Thus, spoken tone2 words that have a tone3 counterpart are lexically ambiguous between a tone2 and a sandhi tone2 (underlyingly tone3). Thus while "yu2" in isolation means fish, in the sequence "yu2 hen3" it might mean either fish or rain. A cross modal priming experiment examined the processing of such potentially ambiguous words in isolation. Visual targets were Chinese characters of four kinds: identical to the auditory word, differentonly- in-tone, irrelevant to the auditory word or nonword. They were presented immediately after auditory primes of four types: tone2 word with tone3 counterpart, tone2 word without tone3 counterpart, tone3 word with tone2 counterpart, or tone3 word without tone3 counterpart. Listeners’ response times were measured as they made word/non-word judgments. RTs were slower for potentially ambiguous words (tone2 words with tone3 counterparts) than for unambiguous words (tone2 words without tone3 counterparts). RTs to tone3 words with or without tone2 counterparts did not differ. The result suggests integration of tone and segmental information during word recognition, without recourse to a "tonal level", which predicts comparable RTs for all tone2s.