A fundamental limitation of gridded precipitation data sets is the paucity of observations over the oceans in the presatellite era. We have evaluated the capabilities and limitations of reconstructing oceanic precipitation using land-based precipitation records. The reconstruction method is based on eigenvector projection, where eigenvectors (or empirical orthogonal functions) of precipitation variability in the Climate Prediction Center (CPC) merged analysis of precipitation (CMAP) gauge/satellite precipitation data set, with near-global coverage from 1979 to present, are projected onto land-only data. The approach has useful skill in areas of the tropics where interannual variability is mainly associated with El Ni¿o--Southern Oscillation (ENSO) activity. Reconstruction is less skillful in non-ENSO years and is primarily restricted to within 1000--2000 km of land-based data. Analogous dependencies hold for the extratropical oceans, where precipitation is best reconstructed close to land data and/or in areas where precipitation variability is locally associated with major atmospheric circulation patterns. Using the National Centers for Environmental Prediction-National Center for Atmospheric Research reanalysis precipitation as a proxy for observations in the presatellite era yields similar results, and in addition, illustrates the limitations to the approach when applied in the presatellite era (prior to 1979): skill is greatest in oceanic areas that are close to land observations and/or dominated by persistent atmospheric or atmosphere-ocean patterns. Overall, the technique would appear to have some value in providing a qualitative picture of oceanic precipitation variability in the presatellite era, but estimation errors preclude use of the data for quantitative analyses.