Aim We investigated partitioning of aquatic macroinvertebrate diversity in eight headwater streams to determine the relative contributions of a and b diversity to c diversity, and the scale dependence of a and b components. Location Great Dividing Range, Victoria, Australia. Methods We used the method of Jost (Ecology, 2007, 88, 2427–2439) to partition c diversity into its a and b components. We undertook the analyses at both reach and catchment scales to explore whether inferences depended on scale of observation. Results We hypothesized that b diversity would make a large contribution to the c diversity of macroinvertebrates in our dendritic riverine landscape, particularly at the larger spatial scale (among catchments) because of limited dispersal among sites and especially among catchments. However, reaches each had relatively high taxon richness and high a diversity, while b diversity made only a small contribution to c diversity at both the reach and catchment scales. Main conclusions Dendritic riverine landscapes have been thought to generate high b diversity as a consequence of limited dispersal and high heterogeneity among individual streams, but this may not hold for all headwater stream systems. Here, a diversity was high and b diversity low, with individual headwater stream reaches each containing a large portion of c diversity. Thus, each stream could be considered to have low irreplaceability since losing the option to use one of these sites in a representative reserve network does not greatly diminish the options available for completing the reserve network. Where limited information on individual taxonomic distributions is available, or time and money for modelling approaches are limited, diversity partitioning may provide a useful ‘first-cut’ for obtaining information about the irreplaceability of individual streams or subcatchments when establishing representative freshwater reserves.