The computer-based College English Test-Spoken English Test (CET-SET) in China incorporates interactional competence in its construct by using a non-face-to-face paired discussion task. To build validity arguments for this unique computer-mediated interactive task, the present study investigated the impact of delivery mode on test performance in terms of communication strategy use. Data were collected from 60 test-takers who sat both the computer-based CET-SET and its face-to-face equivalent. Through an analysis of test-taker discourse combined with post-test test-taker interviews, the study examined their use of strategies in the discussion task under each mode. The findings of the study provided empirical evidence for the comparability of the two modes. Some differences, however, emerged in the use of interaction strategies, which were primarily attributed to the presence/absence of the visual elements of oral interaction. Hence, suggestions were made to address the issues concerning the computer-delivered mode.