Abstract. We compared the functional response of a biogeochemical data assimilation model versus an empirical satellite-derived algorithm in describing the variation of four phytoplankton (diatoms, cyanobacteria, coccolithophores and chlorophytes) groups globally and in 12 major oceanographic basins. Global mean differences of all groups were within ~ 15% of an independent observation data base for both approaches except for satellite-derived chlorophytes. Diatoms and cyanobacteria concentrations were significantly (p < 0.05) correlated with the independent observation data base for both methods. Coccolithophore concentrations were only correlated with the in situ data for the model approach and the chlorophyte concentration was only significantly correlated to the in situ data for the satellite-derived approach. Using monthly means from 1998–2007, the seasonal variation from the satellite-derived approach and model were significantly correlated in 11 regions for diatoms and in 9 for coccolithophores but only in 3 and 2 regions for cyanobacteria and chlorophytes. Most disagreement on the seasonal variation of phytoplankton composition occurred in the North Pacific and Antarctic where, except for diatoms, no significant correlation could be found between the monthly mean concentrations derived from both approaches. In these two regions there was also an overestimate of diatom concentration by the model of ~ 60% whereas the satellite-derived approach was closer to in situ data (8–26% underestimate). Chlorophytes were the group for which both approaches differed most and that was furthest from the in situ data. These results highlight the strengths and weaknesses of both approaches and allow us to make some suggestions to improve our approaches to understanding phytoplankton dynamics and distribution.