Shear velocity was estimated from current measurements near the bottom off Grays Harbor, Washington between May 4 and June 6, 2001 under mostly wave-dominated conditions. A downward-looking pulse-coherent acoustic Doppler profiler (PCADP) and two acoustic-Doppler velocimeters (field version; ADVFs) were deployed on a tripod at 9-m water depth. Measurements from these instruments were used to estimate shear velocity with (1) a modified eddy-correlation (EC) technique, (2) the log-profile (LP) method, and (3) a dissipation-rate method. Although values produced by the three methods agreed reasonably well (within their broad ranges of uncertainty), there were important systematic differences. Estimates from the EC method were generally lowest, followed by those from the inertial-dissipation method. The LP method produced the highest values and the greatest scatter. We show that these results are consistent with boundary-layer theory when sediment-induced stratification is present. The EC method provides the most fundamental estimate of kinematic stress near the bottom, and stratification causes the LP method to overestimate bottom stress. These results remind us that the methods are not equivalent and that comparison among sites and with models should be made carefully. ?? 2006 Elsevier Ltd. All rights reserved.