The present task-based study contributes to the literature on Computer-Assisted Language Acquisition (CALL) by investigating second language (L2) learners' interaction and anxiety while using three different modes of communication: video CMC (VCMC), written CMC (WCMC), and face-to-face (FTF). This investigation has two main goals: first, to explore possible differences in the number of words and turns produced by participants in the three modes while carrying out the tasks and second, to examine learners' state anxiety in the different modalities. Seventy-nine learners from intact intermediate Spanish classes completed three jigsaw tasks in a counterbalanced research design. Participants met on separate occasions in order to carry out three tasks; one task was performed FTF, another using Skype video (i.e., VCMC) and the last using Skype chat (i.e., WCMC). Learners' state anxiety was measured halfway through and upon completion of each task by means of a state anxiety questionnaire. Analyses of the transcripts of the conversations revealed that interaction was similar in terms of number of words and turns during FTF communication and VCMC but significantly lower during WCMC. Results also showed that reported state anxiety was significantly higher during VCMC and FTF than during WCMC. These results are discussed in the context of the relevant literature and their implications drawn for future implementation in the L2 classroom.