Previous studies have shown that adult second language (L2) listeners often experience difficulty encoding language-specific phonological contrasts in word recognition. However, most research on L2 lexical representations has focused on consonants and vowels, and much remains unknown on how lexical tones are encoded in L2 phonological lexicon. In the current study, two experiments were conducted with twenty English learners of Mandarin and 20 Mandarin native speakers. In an ABX task, native speakers outperformed L2 listeners with higher accuracy rate and shorter response latencies. However, both groups showed poor discrimination sensitivity for pairs sharing similar tone contours (i.e., T2-T3). In a medium-lag repetition priming task, listeners were presented with a prime followed by a target that is either the same as the prime (e.g., ni2-ni2) or the other member of a minimal tone pair (e.g., ni2-ni3), eight to 20 items further down in the list. Results show that while significant facilitations in the repetition condition were observed in both groups, in the minimal-tone-pair condition (i.e., T2-T3), positive priming was observed only in the L2 group. The results of the two experiments provide insight into the interface between phonological and lexical levels in L2 spoken word recognition.