The ongoing question of whether negative transfer of speech acts increases or decreases with gains in L2 proficiency has brought mixed results, with some arguing that negative transfer increases with gains in L2 proficiency (Bu, 2012; Morkus, 2018) and others arguing the opposite (Maeshiba, Yoshinaga, Kasper, & Ross, 2006). In response, this study measured the frequency and content of refusals among 33 beginning and 49 intermediate Japanese students of English. Twenty-three English-speaking Americans and 28 Japanese-speakers served as reference groups Data was collected using a discourse completion task developed by Beebe et al. (1990) which included 12 scenarios in which participants must make a refusal to one of four potential stimulus types: 1) refusal to a request, 2) refusal to an invitation, 3) refusal to an offer or 4) refusal to a suggestion. Three levels of interlocutor status were measured according to the participant’s status level: 1) higher, 2) lower, and 3) equal. The frequency of negative transfer was measured using a statistical analysis using a chi-square test for goodness-of-fit suggested that negative transfer increases with L2 proficiency. Intermediates relied on negative transfer in three of the four lower-status scenarios and in all equal-status scenarios. Negative occurred once in the lower-status scenario and once in the higher-status scenario among the beginners. A qualitative analysis revealed that the content of adjuncts was only sensitive to L2 proficiency in equal-status scenarios among the intermediate group. Future research should examine the interconnections between content and the patterning of negative transfer.