During a JGOFS cruise in Dec./Jan. 95/96 various biooptical measurements and experiments on relationships between photosynthesis and irradiance (PAR, UVR) were performed in the Atlantic sector of the Southern Ocean in order to assess its regional variability as defined by provinces and to provide basic data for the estimation of productivity in this region. Data were used for calculating areal primary production and for establishing a diagnostic model of estimating the mesoscale distribution of primary production within the survey area. Measurements of pigment composition by HPLC and size fractionated chl a were used to characterise the phytoplankton communities. For the use of remotely sensed reflectance data (Ri) to obtain pigment concentrations, in situ Ri was compared to surface chl a. Phytoplankton composition and primary productivity in the study showed a regional distribution which was reflected in different biogeochemical provinces of the Southern Ocean: a bloom of large diatoms and very high productivity (>1000 mg C m-2 d-1) characterised the Antarctic Polar Front, phytoplankton of small size, scarce biomass and low primary productivity (<300 mg C m-2 d-1) the interfrontal area of the Antarctic Circumpolar Current, and a Phaeocystis bloom of large colonies and high productivity (560 mg C m-2 d-1) the marginal ice zone. The observed phytoplankton blooms occurred at sites of shallow upper mixed layers (UML; at least <50 m) and stratified water columns creating a light climate giving enough potential to result in high production and growth rates. Also differences in biooptical characteristics coincided with the different biogeochemical provinces of the study: 1. enhanced levels of UVR impaired primary production in the Southern Ocean more strongly outside of phytoplankton blooms than within. In all samples UV-absorbing compounds had been measured, but outside the blooms due to small cell size (<20 µm) efficient protection to UVR by these substances could not be achieved. 2. specific absorption by phytoplankton, af*[l], within the blooms was far lower than outside of the blooms, and decreased significantly below the UML. These differences were due to differences in pigment composition and the package effect, which depends on phytoplankton cell size and photoacclimation in accordance to the UML. 3. The comparison of in situ chl a to in situ Ri of our survey showed that the global-processing-algorithm (Evans & Gordon 1994) and the Southern-Ocean-algorithm (Mitchell & Holm-Hansen 1991) fail for deriving chl a from remote sensing Ri data, because observed regional differences in af*[l] result in different relationships between chl a and Ri . Using specific algorithms for the different provinces improved estimates by 50%. Hence, the spatial discontinuity in biooptical characteristics has to be considered when extrapolating in situ observations relevant for the estimation of primary production in space and time, to match satellite data.