Over the open ocean, the aerodynamic drag coefficient is typically well predicted; however, the impact depth‐limited processes have on the drag remains underexplored. A case study is presented here where winds, waves, and currents were simultaneously observed from a mobile platform that repeatedly transected the inner shelf of Monterey Bay, CA. Eddy covariance‐derived drag coefficients were compared to several bulk parameterizations, including all of the roughness variations of COARE 3.5 and two explicitly depth‐limited models. The analysis demonstrated that the drag was underestimated by O(2–4) times and the variability with wind speed or cross‐shore distance was not well predicted. The drag based on a recent depth‐limited roughness length model performed substantially better than the rest of the bulk estimates, which were all within 15% of each other and effectively equivalent given typical operational uncertainties. The measured friction velocity was compared to a wave‐dependent parameterization and generalizing the model to arbitrary water depth significantly improved the mean observation‐model difference to within 30%. Latent variability in the observation‐model comparison was associated with stability, wind direction, and wave steepness. The wind stress angle variability was also analyzed. Stress veering was correlated with the alongshore surface current within 2 km from shore (r2= 0.7–0.95, p < 0.05); offshore of this margin, consistent wind stress veering was observed and may be attributable to a secondary, low‐frequency swell system. These results demonstrate that it remains a persistent challenge to accurately predict wind stress variability in the nearshore, especially at locations with complex wave and current fields.