Elicited imitation (EI) has gained popularity with recent interests in the quest for efficient and flexible measures of second language (L2) proficiency. Despite the surge of interests, questions remain as to what specific linguistic knowledge, skills, and strategies EI measures. To contribute to this line of inquiry, this study explored the nature of strategy use and its effect on EI performance to elucidate the constructs of EI. Twenty-four L2 learners and eight native speakers of Chinese completed an EI test of Chinese and a strategy use questionnaire after the test. Qualitative analyses of the questionnaire responses revealed that participants mainly employed five types of strategies, including approach strategies, cognitive strategies, metacognitive strategies, communication strategies, and test-wiseness strategies. While native speakers reported the least number of strategies, higher-proficiency L2 learners reported more strategies than lower-proficiency L2 learners. We further subjected strategy use, along with participant proficiency level, item length, and item complexity level, to linear mixed-effects regression analyses. The results showed that participant proficiency level, item length, and item complexity level explained the largest test score variance; in contrast, strategy use of different types only accounted for a smaller proportion. The total number of cognitive strategies had a significant, positive effect on EI performance whereas the total number of metacognitive strategies had a significant, negative effect. These findings offer some insights into the nature of speech comprehension and production on EI and provide validity evidence for the use of EI as a language proficiency measure.