Convective circulation of the mantle causes deflections of the Earth's surface that vary as a function of space and time. Accurate measurements of this dynamic topography are complicated by the need to isolate and remove other sources of elevation, arising from flexure and lithospheric isostasy. The complex architecture of continental lithosphere means that measurement of present-day dynamic topography is more straightforward in the oceanic realm. Here, we present an updated methodology for calculating oceanic residual bathymetry, which is a proxy for dynamic topography. Corrections are applied that account for the effects of sedimentary loading and compaction, for anomalous crustal thickness variations, for subsidence of oceanic lithosphere as a function of age, and for non-hydrostatic geoid height variations. Errors are formally propagated to estimate measurement uncertainties. We apply this methodology to a global database of 1,936 seismic surveys located on oceanic crust and generate 2,297 spot measurements of residual topography, including 1,161 with crustal corrections. The resultant anomalies have amplitudes of ±1 km and wavelengths of ∼1,000 km. Spectral analysis of our database using cross-validation demonstrates that spherical harmonics up to and including degree 30 (i.e. wavelengths down to 1,300 km) are required to accurately represent these observations. Truncation of the expansion at a lower maximum degree erroneously increases the amplitude of inferred long-wavelength dynamic topography. There is a strong correlation between our observations and free-air gravity anomalies, magmatism, ridge seismicity, vertical motions of adjacent rifted margins, and global tomographic models. We infer that shorter wavelength components of the observed pattern of dynamic topography may be attributable to the presence of thermal anomalies within the shallow asthenospheric mantle.