In the Australian arid zone, the species richness of ants is greater and that of mammalian grainvores is less than in North American deserts. This study aimed to determine if the structure of harvester ant communities differs from that seen in North American deserts, focussing on differences related to the paucity of rodents. We tested three hypotheses: (1) because there are fewer rodents, Australian harvester ants should be more abundant and diverse in local habitats than in North American deserts of similar productivity; (2) because the absence of rodents would allow ants to use larger seeds that are preferred by rodents in North America, Australian ant communities should include a larger size range and contain larger workers; and (3) that apart from differences resulting from a paucity of rodents, Australian and North American communities would be convergent in characteristics of community structure resulting from competition. We sampled 19 communities across a climatic gradient in the Australian arid zone and compared the results with data previously obtained for 10 North American communities. Australian harvester ants exhibited similar alpha (within—habitat) diversity but higher beta (between—habitat) diversity between communities. Australian and North American communities were similar in species richness, species diversity, numbers of common species, and abundance of ants, although Australian communities tended to be richer and more diverse at sites with lower precipitation. North American communities increased in species richness, diversity, numbers of common species, and proportions of column—foraging species with increasing precipitation. In contrast, Australian communities did not change regularly across a gradient in precipitation. Thus, the first hypothesis was rejected. The second hypothesis was also rejected because Australian harvester ants were smaller, covered a smaller size range, and tended to be more tightly distributed along the size gradient. Dietary data indicated a lack of correspondence between resource use and availability, suggesting that Australian communities may not generally be in equilibrium with their resource environment. However, neither of the first two hypotheses could be tested unequivocally because it remains possible that evolutionary interaction between seeds and ants in the absence of rodents has allowed seeds to adopt defenses minimizing consumption by ants, or that other granivores (particularly birds) compensate in part for the paucity of rodents. Tests of the third hypothesis were ambiguous. Some results suggest basic similarities between the effects of competition on the communities, such as the similar maximum values for abundance, richness, and species diversity. However, several attributes of Australian communities differed from those in North America: foraging occurred over a wider range of soil temperatures; temporal displacement of foraging among coexisting species was prominent; and individually foraging species occurred with equal frequency across climatic gradients. We discuss various biotic and abiotic features that may explain differences in community structure between continents. We conclude that the paucity of rodents, a potentially competing group of granivores, has not led to predictable changes in harvester ant communities of arid Australia.