The latitudinal distributions of phytoplankton biomass, composition and production in the Atlantic Ocean were determined along a 10,000-km transect from 503N to 503S in October 1995, May 1996 and October 1996. Highest levels of euphotic layer-integrated chlorophyll a (Chl a) concentration (75}125 mg Chl m\ ) were found in North Atlantic temperate waters and in the upwelling region o! NW Africa, whereas typical Chl a concentrations in oligotrophic waters ranged from 20 to 40 mg Chl m\ . The estimated concentration of surface phytoplankton carbon (C) biomass was 5}15 mg C m\ in the oligotrophic regions and increased over 40 mg C m\ in richer areas. The deep chlorophyll maximum did not seem to constitute a biomass or productivity maximum, but resulted mainly from an increase in the Chl a to C ratio and represented a relatively small contribution to total integrated productivity. Primary production rates varied from 50 mg C m\ d\ at the central gyres to 500}1000 mg C m\ d\ in upwelling and higher latitude regions, where faster growth rates ( ) of phytoplankton ( '0.5 d\ ) were also measured. In oligotrophic waters, microalgal growth was consistently slow [surface averaged 0.21$0.02 d\ (mean$SE)], representing (20% of maximum expected growth. These results argue against the view that the subtropical gyres are characterized by high phytoplankton turnover rates. The latitudinal variations in were inversely correlated to the changes in the depth of the nitracline and positively correlated to those of the integrated nitrate concentration, supporting the case for the role of nutrients in controlling the large-scale distribution of phytoplankton growth rates. We observed a large degree of temporal 0967-0637/00/$ -see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 7 -0 6 3 7 ( 9 9 ) 0 0 0 8 7 -4 variability in the phytoplankton dynamics in the oligotrophic regions: productivity and growth rates varied in excess of 8-fold, whereas microalgal biomass remained relatively constant. The observed spatial and temporal variability in the biomass speci"c rate of photosynthesis is at least three times larger than currently assumed in most satellite-based models of global productivity.