This paper investigates the mediation role of interpreters through modality shifts in Chinese-English diplomatic interpreting. Based on the custom-built corpus of interpreted government press conferences, this article conducts a Systemic Functional Linguistics-informed analysis of modality shifts by examining the ST-TT sentence pairs that present high-frequency Chinese or English modality markers. Results show that: (1) the degree of mediation by the interpreters is fairly high in terms of modality, with 44% of the investigated sentence pairs involving modality shifts; (2) shifts mostly occur in modality value (91%) and orientation (64%) rather than modality type (5%), thus the basic speech function of the clause is minimally changed; (3) shifts within the three dimensions point to general tendencies towards “weakening,” “subjectivization” and “de-obligation,” which reflect the interpreters’ efforts to construct the Chinese officials’ image as audience-friendly and proactive, and to adapt the TT to the English communicative norms; (4) “strengthening” and “objectivization” shifts occur frequently around issues concerning the Chinese government’s responsibility or ability, which reveals the interpreters’ active involvement in presenting the Chinese government as responsible, confident and powerful. The study argues that the press conference interpreters’ active mediation is motivated by their institutional identity as “diplomatic workers” and “government representatives” in the Chinese context.