Aim: The 'taxonomic impediment' has led to a growing trend in ecology and conservation biology to use operational surrogates for species within the context of a particular research project. Because such 'parataxonomic' classifications are typically spatially limited in scope, we examined the influence of increasing spatial scale on the congruence of two such approaches with a more traditional taxonomic classification. Location: Sturt National Park, north-western New South Wales, Australia. Methods: Specimens of two ant genera, Camponotus and Rhytidoponera, were classified by three independent methods. The 'parataxonomic' classification assigned specimens to morphospecies without specialist taxonomic expertise; the 'taxonomic' classification assigned specimens to either described species or, where this was not possible, to operational taxonomic units (OTUs) using specialist taxonomic expertise; the 'phenetic' classification assigned specimens to putative species using a K-means partitioning algorithm on basic morphometric data. Specimens of each genus were pooled into 'assemblages', which were defined at multiple spatial scales using a nested sampling design. Congruence in the interspecimen relationships of the different classifications was tested for each assemblage using pair-wise Mantel correlations. Results: Classification congruence tended to decrease with increasing spatial scale. There were, however, clear differences between the genera. Parataxonomic-taxonomic congruence was consistently greater for Camponotus, while phenetic-taxonomic congruence showed the opposite pattern. Conclusions: Observed patterns in classification congruence are attributed to two principal causes: (i) within-species morphological variation, including ecotypic variation in Rhytidoponera and caste polymorphism in Camponotus; and (ii) a limit to the morphological similarity of potentially competing species at small spatial scales. Regardless of cause, the decline in agreement as the spatial scale of observation is increased has important implications for the measurement of biodiversity, particularly when comparing samples over regional, continental, and global scales.