Background: An optimal habitat selection model called centrifugal community organization (CCO) predicts that species, although they have the same primary habitat, may co-exist owing to their ability to use different secondary habitats. Goal: Test the predictions of CCO with field experiments. Species: The Egyptian sand gerbil (40 g), Gerbillus pyramidum, and Allenby's gerbil (25 g), G. andersoni allenbyi. Site: Ashdod sand dunes in the southern coastal plain of Israel. Three sandy habitats are present: shifting, semi-stabilized, and stabilized sand. Gerbils occupied all three habitats. Methods: We surveyed rodent abundance, activity levels, and foraging behaviour while experimentally removing G. pyramidum. Results: Three predictions of the CCO model were supported. Both species did best in the semi-stabilized habitat. However, they differed in their secondary habitats. Gerbillus pyramidum preferred the shifting sand habitat, whereas G. a. allenbyi preferred the stabilized habitat. Habitat selection by both species depended on density. However, in contrast to CCO, G. pyramidum dominated the core habitat and excluded G. a. allenbyi. We term this variant of CCO, 'asymmetric CCO'. Conclusions: The fundamental feature of CCO appears valid: co-existence may result not because of what each competing species does best, but because of what they do as a back-up. But in contrast to the prediction of the original CCO model, all dynamic traces of interaction can vanish if the system includes interference competition. ?? 2007 Gideon Wasserberg.