Several aspects of feedback mechanisms associated with surf zone sandbar response have been characterized using bathymetric surveys, sampled approximately monthly over a 16-year period at the Army Corps of Engineers' Field Research Facility (North Carolina). The measured bathymetry was alongshore averaged and modeled by the superposition of two Gaussian-shaped sandbars on an underlying planar slope. A third, half-Gaussian-shaped bar represented steepening at the shoreline. The rms error between the measured bathymetry and the profile model was 0.10 m (estimated over 322 different surveys). The model explained 99% of the profile variance that remained after first removing the linear, cross-shore trend from each observed profile. Bar response, which was extracted from the modeled profiles, was compared to a local hydrodynamic forcing variable &Ggr; (&Ggr; was defined as the ratio of the wave height to water depth, evaluated at bar crest locations). At low values of &Ggr; (i.e., nonbreaking conditions), bars migrated onshore, and their amplitude tended to decay. At high values of &Ggr; (i.e., breaking conditions), bars migrated offshore, with relatively little change in amplitude. The transition between onshore and offshore migration occurred at a value of &Ggr; that was consistent with the onset of wave breaking. Bar migration was associated with a stabilizing feedback mechanism, which drove bar crests toward an equilibrium position at the wave breakpoint. However, we observed that the rate of bar response showed no reduction for any nonzero choice of &Ggr;, indicating that bars never reached equilibrium. Systematic bar amplitude decay was observed under nonbreaking conditions. Bar amplitude decay could drive &Ggr; farther away from breaking conditions, allowing further bar amplitude decay. This is a destabilizing feedback mechanism, potentially leading to bar destruction. ¿ 2001 American Geophysical Union