Fairy circles (FCs) are extremely ordered round patches of bare soil within arid grasslands of southwestern Africa and northwestern Australia. Their origin is disputed because biotic factors such as insects or abiotic factors such as edaphic and eco-hydrological feedback mechanisms have been suggested to be causal. In this research, we used a multi-scale approach to shed light on the origin of Australian FCs. At a local scale, we investigated the potential cause of FCs using analyses of soil compaction and texture within FCs, the surrounding matrix vegetation, and in nearby large bare-soil areas. We found that soil hardness and clay content were similarly higher inside the FCs and in the large bare-soil areas. When compared to the matrix soils with protective grass cover, the 2.6–2.8 times higher clay content in FCs and large bare-soil areas is likely sourced via multiple abiotic weathering processes. Intense rainfall events, particle dispersion, surface heat, evaporation, and mechanical crust building inhibit plant growth in both areas. At the landscape scale, a systematic survey of 154 soil excavations within FCs was undertaken to evaluate the presence of pavement termitaria that could inhibit plant growth. We show that in up to 100% and most of the excavations per plot, no hard pavement termitaria were present in the FCs. This fact is substantiated by the observation that small, newly forming FCs are initiated on soft sand without evidence of termite activity. At the regional scale, we investigated the spatial properties of FCs and common termite-created gaps in Western Australia, using spatially explicit statistics. We mapped three 25-ha FC plots with a drone and compared them with three aerial images of typical vegetation gaps created by harvester and spinifex termites. We demonstrate that the small diameters, the lower ordering, and the heterogeneous patterns of these common termite gaps strongly differ from the unique spatial signature of FCs. Our multi-scale approach emphasizes that FCs are not trivial termite gaps and that partial correlation with termites at some sites does not imply causation. Instead, we highlight the need to study the edaphic and eco-hydrological drivers of vegetation-pattern formation in water-limited environments.