Global estimates of sea-surface temperature (SST) distributions represent an important prescribed boundary condition for atmosphere-only general circulation models (AGCMs) which aim to Simulate the behaviour of the past climate system. We investigate the use of objective interpolation techniques, based on Delaunay triangulation and gridded-averaging, in translating point-based SST data into global maps. Existing global SST datasets generated by CLIMAP for the Last Glacial Maximum (LGM) and PRISM for the middle Pliocene are used as vehicles for evaluation. Results indicate that the sample quantity and array of the CLIMAP sites are sufficient to faithfully reconstruct major surface water patterns of the LGM. However, as problems with sample quantity and distribution become more acute in pre-Quaternary periods, interpolators are limited in their ability to produce reasonable estimates in data sparse regions. In such cases, potential biases associated with interpolation become significant enough to bring into doubt the overall validity of the global SST reconstruction. This limits the datasets use as a prescribed boundary condition in AGCM simulations. On a regional scale, objective methods offer a valuable means of assessing data quality and constraining future Subjective analysis. Primary rather than interpolated SST data remain extremely valuable as a tool to evaluate output from coupled ocean-atmosphere models rather than a prescribed boundary condition in AGCM Simulations.