Patterns of phytoplankton in areas with seagrass (Zostera capricorni, Halophila ovalis, and Ruppia megacarpa) and areas without seagrass were quantified in an estuary near Sydney, NSW, Australia. Replicate samples (separated by metres) were collected on 20 randomly chosen occasions over almost 3 years, from two sites (separated by tens of metres) nested in both habitats, at four locations (separated by kilometres). Areas with seagrass often supported a significantly different assemblage of phytoplankton compared with areas without seagrass. Dissimilarities were due to small differences in the relative abundance of taxa, particularly the diatom Pseudonitzschia cf. pseudodelicatissima, rather than the presence or absence of specific taxa. In both habitats, abundances of phytoplankton were patchy at all of the spatial scales sampled, although most of the variation was among replicate samples. These findings emphasise the need to incorporate a hierarchy of scales in programs designed to understand processes that cause spatial and temporal patterns of phytoplankton or to detect changes in response to human activities.