We present a data set of 738 planktonic foraminiferal species counts from sediment surface samples of the eastern North Atlantic and the South Atlantic between 87¿N and 40¿S, 35¿E and 60¿W including published Climate: Long-Range Investigation, Mapping, and Prediction (CLIMAP) data. These species counts are linked to Levitus's <1982> modern water temperature data for the four caloric seasons, four depth ranges (0, 30, 50, and 75 m), and the combined means of those depth ranges. The relation between planktonic foraminiferal assemblages and sea surface temperature (SST) data is estimated using the newly developed SIMMAX technique, which is an acronym for a modern analog technique (MAT) with a similarity index, based on (1) the scalar product of the normalized faunal percentages and (2) a weighting procedure of the modern analog's SSTs according to the inverse geographical distances of the most similar samples. Compared to the classical CLIMAP transfer technique and conventional MAT techniques, SIMMAX provides a more confident reconstruction of paleo-SSTs (correlation coefficient is 0.994 for the caloric winter and 0.993 for caloric summer). The standard deviation of the residuals is 0.90¿C for caloric winter and 0.96¿C for caloric summer at 0-m water depth. The SST estimates reach optimum stability (standard deviation of the residuals is 0.88¿C) at the average 0- to 75-m water depth. Our extensive database provides SST estimates over a range of -1.4 to 27.2¿C for caloric winter and 0.4 to 28.6¿C for caloric summer, allowing SST estimates which are especially valuable for the high-latitude Atlantic during glacial times. ¿ American Geophysical Union 1996