We closely replicated Wu and Ortega (2013), who found that an elicited imitation test (EIT) reliably distinguished low-level from high-level language abilities among instructed second language (L2) learners of Mandarin Chinese. The original study sampled learners: (1) from second-level courses to represent low-level language abilities; and (2) from third-, fourth- and graduate-level courses to represent high-level language abilities. Results showed high-level learners outperformed low-level learners on the Mandarin EIT. Our close replication used Wu and Ortega's (2013) materials and procedures in order to understand: (1) the extent to which this EIT can additionally distinguish between finer-grained language abilities; and (2) the ways in which the broad grouping of language abilities in the high group may have contributed to the findings. Sixty-five instructed L2 learners from four instructional levels were assigned to one of three groups: Beginner (first-level courses), Low (second-level courses), High (third- and fourth-level courses). Consistent with the original study, our results showed clear between-group differences, indicating that the EIT can distinguish between both broad (beginner vs high) and finer-grained (beginner vs low, low vs high) language abilities. These results are discussed in light of the original study's findings with implications for proficiency assessment in second language acquisition (SLA) research.