Measurements of seabed drag coefficient, C100, were made under tidal currents at four sites in Mahurangi Harbour, New Zealand. At the first three sites the dominant roughness element was the pinnid bivalve, Atrina zelandica (horse mussel). At the fourth site, which was devoid of horse mussels but covered in cockle shells, patches of seaweed and crab burrows, C100 was smallest (0.0055), but still twice as large as the value typically applied to abiotic, flat, cohesionless seabeds (0.0025). The mean drag coefficient plus-or-minus standard error at the three sites with horse mussels was: 0.0082 ± 0.0010 (site 1); 0.0096 ± 0.0009 (site 2); 0.0115 ± 0.0016 (site 3). There were no clear differences amongst sites 1, 2 and 3 in terms of the attributes of individual horse mussels (e.g. shell height, width or orientation), which could have been used to explain the ranking of the drag coefficients. There were, however, differences amongst the three sites in terms of spatial distribution of individual bivalves. The site with the highest density of horse mussels, site 1, had the lowest drag coefficient and an areal concentration (λ) of horse mussels higher than typical values cited for the critical concentration (λc) for the onset of skimming flow over various idealized, three-dimensional roughness elements. At sites 2 and 3, the drag coefficient was given by: C100=[κ/1n (300/mkλ)]2 which was valid for λ < λ c, where κ is von Karman's constant, k is the horse mussel height (i.e., protrusion above the seabed), m ≈ 100 and λc ≈ 0.2. The stable eddies that are hypothesized to lodge between roughness elements at concentrations greater than λc may influence benthic community dynamics.