This dataset contains 574 household-level observations collected to assess community perceptions and local mitigation practices related to mazuku—areas of hazardous carbon dioxide diffuse degassing—in the western sector of Goma, located in the Virunga Volcanic Province (DR Congo). The data were generated through a mixed-methods design combining quantitative household surveys and qualitative interviews, conducted across three sampling zones representing different levels of exposure to CO₂ emissions. The dataset documents demographic characteristics, household experiences with volcanic and mazuku-related hazards, and detailed responses on the perceived implementation and perceived effectiveness of three locally recognised categories of mazuku mitigation measures: (1) emission-limiting measures (e.g., blocking gas pathways with waste materials), (2) adaptive measures (e.g., ventilation practices or occupying upper floors), and (3) awareness-based measures derived from orally transmitted local knowledge (e.g., avoiding mazuku areas early in the morning). Variables also capture socio-economic factors—such as income, gender, and duration of residence—that influence motivation and perceived efficacy of mitigation strategies. Spatial identifiers are included to allow comparison between sampling zones and enable analyses of geographical patterns in implementation and perceived effectiveness. This dataset provides a unique empirical foundation for understanding community-driven risk reduction practices associated with carbon dioxide diffuse degassing hazards in a rapidly urbanising Global South context. It is intended to support further interdisciplinary research on volcanic gas hazards, inform co-designed mitigation strategies with local communities, and encourage future integration of physical monitoring of mazuku concentrations with social perception data.