Large clastic dykes (the Harutori‐Taro and Harutori‐Jiro dykes) and smaller dykes are exposed in the underground Kushiro Coal Mine (KCM), Japan. This study examines these dykes as a case study to investigate the geological conditions and fluid flow history that lead to the development of large clastic dykes in basins. The composition of the dykes indicates the Beppo and/or Harutori formations as their parent unit. Crystallite size distribution (CSD) analysis reveals Ostwald ripening of the kaolinite in the kaolinitised feldspar from the dykes, suggesting stagnant conditions in the parent unit before the dyke was formed. In contrast, smectite CSDs and the high carbonate content of the dykes suggest that large volumes of fluid flowed through the dykes along the established hydraulic gradient, which was triggered by the breaking of the upper seal. The isotopic and chemical compositions of the calcite and aragonite in the dykes, with moderate siderite and rhodochrosite content, indicate the fluid was a warm (>30°C) mixture of freshwater and saltwater, which was transferred from deeper levels of the parent unit towards the crest of an anticline. Immediately after sand injection, the semi‐closed system of the parent unit near the root of the large dyke was transformed into a major flow channel for overpressurised fluids. Subsequently, a large volume of fluid flowed along the vertical conduit (or dyke) over a long period of time (>1 Myr), which removed fluid from a widespread area (i.e., several hundred square kilometres) of the basin. The results show that thin parent units, poor lateral continuity of the upper seal, and spatially heterogeneous overpressurisation do not preclude the formation of large dykes.