Mini-fiction is a modern prose that is present on Twitter with its uniqueness on the topics that are thrown at the readers. In fact, not all readers who provide comments follow the systematic writing of mini-fiction, whereas in terms of discourse integration, omission or differences in the title and the topic can be a problem. This study aims to analyze the coherence of comments given by readers on the topic of mini-fiction. This study uses a qualitative descriptive study using a critical discourse paradigm. The data source used was the @fiksimini account on Twitter during March 2020. Data collection was carried out using reading techniques and note-taking techniques in data cards. The data collection instrument is the researcher himself (human instrument) using the coherence relation parameter. The data analysis technique in this study used semantic and pragmatic equivalents in analyzing reader comments based on their coherence relationships. The data analysis stage was carried out in three stages, namely ( 1) word reduction; (2) data presentation; and (3) drawing conclusions. The results showed that the coherence between the topic of and readers' comments was 65.6%. The coherence relation tools found in the discourse are divided into (1) causation; (2) equations; (3) options; (4) conflict; (5) temporal; (6) objectives; (7) conditions; (8) comparison; (9) affirmation; (10) sequence; and (11) the situation.