If AUV survey design does not consider the practical limits to change detection and include best practices to minimize noise, however, the results may be ambiguous or even useless. Using a series of simulations combining real bathymetric digital elevation models (DEMs) with idealized perturbations representing seafloor subsidence and faulting, this paper examines limits to seafloor change detection using repeat MBES surveys in terms of the standard deviation of the noise in the surveys. In general, the magnitude of vertical change between surveys should be greater than the standard deviation of the noise (also known as uncertainty or error) in each of the surveys compared in order reliably recognize seafloor change. Filtering can suppress noise and recover the real seafloor change, possibly even if the magnitude of the change is less than the standard deviation of the noise. Based upon the simulations in this paper, filtering is more effective for broad features like subsidence bowls than localized features like faults even if the amount of vertical change is identical.