Headwater streams dominate stream channel length in catchments. They are important sources of water, sediment and biota for downstream reaches and critical sites for organic matter and nutrient processing. Aquatic biodiversity in headwater streams has been overlooked in comparison to higher-order rivers, and few studies have considered spatial biodiversity patterns in headwater streams, or streams in general. I reviewed studies of macroinvertebrate diversity in headwater streams and found equivocal evidence to support the view that headwater streams harbour high biodiversity. Headwater streams might still make an important contribution to [gamma] (regional) diversity at the landscape (catchment) scale by virtue of high [beta] (among-assemblage) diversity. I studied eight headwater streams from three forested, upland catchments along the Great Dividing Range, Victoria, Australia to test my hypothesis of high [beta] diversity and to understand the spatial patterns and determinants of macroinvertebrate diversity in headwater stream networks.Diversity partitioning showed that reaches each had high [alpha] (within-assemblage) diversity, while [beta] diversity made only a small contribution to [gamma] diversity at both the reach and catchment scales. [beta] diversity may have been lower than hypothesized due to relatively small distances between sites and high levels of dispersal among reaches and catchments in the study area. Contrary to other studies that have found environmental factors to be important for explaining variation in macroinvertebrate assemblage structure in headwater streams, I found a limited role for environmental factors structuring macroinvertebrate assemblages in the study area.In one year (2008), spatial factors (independent of environmental factors) were the dominant factor structuring macroinvertebrate assemblages. Therefore, metacommunity structure in the study area aligns most closely with the neutral/patch dynamic metacommunity model. This pattern of spatial structuring, coupled with low [beta] diversity, suggests that high neighbourhood dispersal might be the main factor structuring macroinvertebrate assemblages in the study area. Flow permanence had only a seasonal effect on macroinvertebrate diversity and so there is a temporal component to the spatial diversity patterns in this system.The explicit recognition of stream ecosystems as spatially structured networks has increased our understanding of ecological patterns and processes, and provided the impetus for this research. Recent advances in the study of networks, particularly in the fields of physics and network theory, offer an opportunity to considerably extend the current application of the network concept in stream ecology.Determining the relative contributions of [alpha] and [beta] diversity to [gamma] diversity, and the scale dependence of [alpha] and [beta] components, provides vital information for conservation planning because optimal reserve designs will differ depending on the relative contributions of [alpha] and [beta] diversity. My finding of high [alpha] and low [beta] diversity indicates that each stream in the study area can be considered to have low irreplaceability and the capacity to contribute a large portion of species to regional conservation targets.Information on spatial patterns of diversity is urgently required for systematic conservation planning for freshwater reserves if we are to halt the rapid decline in global freshwater biodiversity.