Anonymous peer feedback has long been used as a common activity in higher education, yet few studies have systematically tracked students' perceived comfort levels over time, and even fewer have compared givers' and receivers' comfort under this intervention. To address this gap, the present study tracked 40 Chinese university students over eight weeks (four weeks' non-anonymous and four weeks' anonymous condition). Based on quantitative data (comfort ratings) and qualitative data (feedback texts and survey responses), it found that students overwhelmingly favoured the anonymous mode, reporting significantly higher comfort ratings at the peer and class-level dimensions. Additionally, they assigned significantly higher comfort values to the giver role where they could expose and exchange more private feelings with peers as if they were in a masked ball. Pedagogical implications are given to inform and improve feedback activities.