This paper presents a comprehensive set of velocity and suspended sediment observations in the nearshore wave bottom boundary layer, collected during the Duck94 field experiment on the Outer Banks of the North Carolina coast. Cross-shore velocity measurements in the wave bottom boundary layer were made using five hot film anemometers, nominally spaced from 1 to 5 cm above the bed in 2 m of water depth. The time-varying location of the seabed was estimated to roughly 1 cm with a stacked set of bed-penetrating fiber-optic backscatter sensors. The instrument array was intermittently located in the surf zone on the crest of a bar. The location of the bottom varied several centimeters over a 34 min data run. Even over 4 min segments of quasi-steady statistics, occasional large waves caused short erosion and redeposition events, complicating the definition of bottom location and causing the root-mean-square velocity statistics to be nonzero below the mean bed location. This leads to obvious difficulties in comparisons with two, one-dimensional time-dependent, eddy viscosity wave bottom boundary layer models. For example, bed shears based on rms amplitude decay were lower than predicted. The observations show some evidence for a velocity overshoot region within the wave bottom boundary layer. The observations were compared with two linear eddy viscosity models. Larger estimates of a constant eddy viscosity and smaller than predicted phase leads are indicative of more rapid mixing of momentum than predicted by the models. The phase and amplitude frequency response estimated with frequency domain empirical orthogonal functions shows a nonlinear response of the wave bottom boundary layer over the incident band. These observations are among the first coherent looks at the wave bottom boundary layer under conditions of significant sediment response. They highlight the added complexity of the dynamics in natural environments. ¿ 2000 American Geophysical Union