The circular 625 km2 alkaline Pilanesberg Complex, South Africa, contains coeval eruptive and several distinctive intrusive syenitic and foyaitic components, concentrically arranged at the surface. However, owing to poor outcrop the relationships between the different intrusive rocks, and their shape in the third dimension cannot be convincingly determined in the field. The original interpretation was a laccolith, whereas later models suggested a funnel shape, and appealed to ring-dyke and cone-sheet emplacement mechanisms. However, the radial widths of these coarse-grained bodies are over 1 km and so cannot have been emplaced as ring dykes or cone sheets, which are usually quite thin and fine grained. Creating the space for emplacement and removal of pre-existing country rocks for each postulated subsequent intrusive event presents a major challenge to this latter hypothesis. Extensive previously published and new field relationships are re-evaluated here to suggest that the body is a gently inward-dipping sheet and that subsequent injections of magma merely pumped up an existing and evolving magma chamber rather than intruded into solid rocks. A Bouguer gravity anomaly model is presented that supports the concept of a shallow, flat-bottomed body rather than one that continues to significant depth. There are many analogies with the Kangerlussuaq Intrusion, Greenland.