The role of physical oceanographical, geochemical, and sedimentological data in the problem of estimating ocean primary productivity is analyzed using geostatistical and algebraic multivariate spatial methods. The available maps mirror difficulties in measuring productivity directly and quantifying biological observations, which result in a very spotty survey coverage of the world's oceans. This paper takes an approach of estimating the target productivity from related sedimentological, physical oceanographical, and marine geochemical variables. The variables are considered as a multivariate spatial system. Geostatistical estimation is applied to fill in the survey gaps, and the individual data sets are then integrated into a multivariate spatial model using algebraic map comparison. Results are used in the quality assessment and calibration of transform models between the proxy variables and primary productivity. For the Atlantic Ocean, case studies are carried out for the phosphate distribution at the 100 m level, foraminifera abundance in sediments, and sea surface temperature. The utility of each proxy variable in the prediction of productivity is discussed. A new map of phosphate concentration at the 100 m level with full coverage of the Atlantic Ocean is complied by application of geostatistics to an enlarged data base containing new observations. Primary productivity can be predicted from this phosphate map, using transforms that involve also distance from the coastline, and latitude (photosynthesis restriction). Foraminifera abundance is in principle closely related to productivity and is of importance in view of a paleoceanographic reconstruction, but difficulties in measuring and quantifying reflected in the quality of the data set prohibit a direct estimation. The relationship between higher fertility and lower temperature known from upwelling areas does not provide a simple global predictor; instead, sea surface temperature seasonability is an indicator of primary productivity under the constraint of regional availability of nutrients. ¿ American Geophysical Union 1992