This article tries to expand the exploration of Cantonese sentence final particles (SFPs), and examines especially their roles and functions in modulating the speaker's epistemic stance in conversational interactions. A speaker's stance is appearing and constantly being treated and molded in the course of the conversation. Given their position as utterance final, SFPs are perfect grammatical devices to be used to recalibrate and shape the speaker's epistemic stance. Considering epistemic modulation as a discourse process, this article elucidates how Cantonese SFPs can be utilized to reaffirm, as well as to modulate (i.e. upgrade or downgrade) the epistemic stance of the speaker in the conversational data. • It is hoped that this paper could also shed light on other cross-linguistic studies on epistemic modulations and stance-taking. Abstract This paper attempts to extend the investigation of Cantonese sentence final particles (SFPs), and explore in particular their roles and functions in modulating the speaker's epistemic stance in conversational interactions. The stance of a speaker is emerging and continuously being negotiated and shaped in the course of the conversation; conversation participants will need to constantly modify and revise their stance and what they have already said as the conversation progresses. Because of their position as utterance final, SFPs are perfect grammatical devices to be employed to recalibrate and finalize the speaker's epistemic stance. Seeing epistemic modulation as a discourse process, this paper exemplifies how Cantonese SFPs can be deployed to reaffirm, as well as to modulate (i.e. upgrade or downgrade) the epistemic stance of the speaker in our conversational data. It is hoped that this paper could also shed light on other cross-linguistic studies on epistemic modulations and stance-taking.