A volunteer observing ship (VOS)-expendable bathythermograph (XBT) network has been proposed for the Atlantic Ocean to satisfy World Ocean Circulation Experiment (WOCE) objectives in the upper water column. These objectives include measuring changes in upper-layer temperature. An evaluation of the proposed WOCE XBT network to resolve variability in sea surface temperature (SST), temperature distribution at 150 m (T150), and average temperature of the upper 400 m (T400L) between 25°S and 35°N is performed. A sampling design study based on an optimum interpolation (OI) of the historical XBT dataset is used to construct uncertainty distributions for various XBT networks. The OI technique requires statistical representations of the variability (in the form of structure functions) of the three variables that are derived from the historical database. The structure functions and various sampling grids are used to construct uncertainty maps. Two seasons are used in the analysis of SST. In both seasons, uncertainties in mapped SST values for the proposed WOCE grid range from 0.3° to 0.4°C in regions of adequate data coverage. Errors are larger along the boundary. Uncertainties in the T150 fields are larger (0.5°−0.7°C) because of the smaller scales of spatial variability at depth. Errors in T400L range from 0.3° to 0.4°C. The effect of alternative observing strategies on the error maps are shown. Finally, error maps derived from the XBT network as it exists today (i.e., incomplete) are given. The maps indicate that monthly resolution is not available from the incomplete network.