Abstract Four bottom roughness models are tested using field data from the inner shelf of the Middle Atlantic Bight. Bottom roughness plays a significant role in calculations of sediment concentration profiles and current velocity profiles. The importance of each of the three parts in the roughness models (grain roughness, ripple roughness and sediment motion roughness) vary depending on forcing conditions. Consistent with the observations of others e.g. Caechione and Drake, 1990 ( The sea, Vol. 9, pp. 729–773); Wiberg and Harris, 1994 ( Journal of Geophysical Research, 99(C1), 775–7879), our results show that the models of Smith and McLean (1977; Journal of Geophysical Research, 82, 1735–1746), Grant and Madsen (1982; Journal of Geophysical Research, 87, 469–481) and Nielsen (1983; Coastal Engineering, 7, 233–257) overestimate the sediment transport roughness under sheet-flow conditions. However, the Nielsen (1983) model can predict the ripple roughness under moderate energy conditions quite well. A refined bottom roughness model is proposed that combines Nielsen's ripple roughness model and a modified sediment motion roughness modelk b=d + 8ννα+ Ωd(ψ′ m − ψ c This sediment motion roughness is defined in such a way that it is proportional to the maximum skin friction Shields' parameter. The proportionality constant, Ω, is determined by fitting the modeled roughnesses and shear velocities with the field observations. The calculated velocity profiles and roughness using the refined roughness model, with Ω = 5, compare well to the field observations made under both moderate and high energy conditions at a sandy inner shelf site.