This study addressed the question 'Does seagrass support a greater diversity and abundance of fish than bare sand over a large spatial scale?' The answer varied with both the measure used to describe the assemblages of fish, and the distance sand was from seagrass. Significantly more species were collected from Zostera capricomi than from sand at least 100 m from this seagrass in 89 % of locations. At 58 O/ O of locations, Zostera also contained more species than sand within 10 m of the seagrass. When 2 species with very patchy distributions were excluded from analyses, Zostera held sigllificantly more fish individuals than sand > 100 m away at 66 % of locations. However, Zostera did not consistently contain more fish individuals than sand within 10 m of seagrass: Zostera had significantly more fish In 25 % of locations, adjacent sand had more fish in 25 % of locations and there were no differences in the remaining comparisons. Most of the common species were found predominantly in Zostera or the adjacent sand. There were prominent dissimilarities in relative abundance of species in samples from the 3 habitats, although samples from Zostera were more similar to those from adjacent sand than to those from sand > 100 m away. We conclude that (1) Zostera supported more species and individuals, and different species, than sand > 100 m from seagrass at a majonty of locations, and (2) the fish assemblage associated with sand within 10 m of Zostera had fewer, and ddferent, species than the seagrass at a majority of locations but supported as many individual fish as Zostera.