Few studies have compared the nature of negotiation of meaning across face-to-face (FTF) and synchronous computer-mediated communication (SCMC) modes with mixed-proficiency dyads. Such comparisons can help identify important strengths and weaknesses of each mode. Therefore, this study explored whether FTF and SCMC modes differ in terms of frequency and quality of negotiations of meaning in English as a foreign language (EFL) learners' interaction in mixed-proficiency dyads. To this end, 32 EFL learners at elementary and upper-intermediate levels were assigned to 16 mixed-proficiency dyads to perform two information gap tasks (i.e., spot the differences task and jigsaw task). The participants' interactions in both modes were coded based on Varonis and Gass's (1985) model of negotiation routines and Long's (1980) negotiation moves. The participants' interactions in each mode were coded. The results show no significant differences between the frequency of negotiations in both modes. However, the results were controversial regarding the quality of negotiations. The quality of negotiation routines was examined by the number of indicators leading to modified output, which revealed that mode of interaction influenced the quality of negotiations and more modified output was produced in the SCMC mode than the FTF mode.