More and more studies emphasize the status of phosphorus (P) as the principal limiting nutrient of phytoplankton growth, especially in coastal waters under the influence of freshwater discharges. The purpose of the present paper is to investigate the role of P on planktonic production in the waters influenced by the Gironde discharges; the Gironde being one of the two largest rivers on the French Atlantic coast. The survey is based on several cruises made in 1998 and 1999. Two different patterns were observed for waters with salinity below and above 34.5. For waters with salinity <34.5. P was found to be the first limiting nutrient of winter and spring phytoplankton blooms, based on undetectable phosphate (<20 nM), high NO3 : PO4 ratios, typically >100:1, short phosphate turnover time (1 to 2 h), high alkaline phosphatase activities (mean of 176 nM h(-1) in late May 1999) and ultimately great increases of chlorophyll a (Chl a) and primary production in phosphate-enriched samples relative to controls. This limitation could be partly explained by the Gironde nutrient supplies, which were phosphate deficient compared with the mineral nitrogen (N-min : PO4 was >40 within a salinity range 16-33). In summer, corresponding to the period of low influence of Gironde supplies (low runoff and a spreading effect of the plume), phytoplankton growth would be controlled by both P and nitrogen (N), according to low nitrate and the major effect of combined P+N (NH4) enrichment on Chl a and primary production compared with the addition of Nor P singly. In early October, after the first autumn gales, the mixed layer was enriched with a sufficient supply of nutrients to support exponential phytoplankton growth for 4 days in enclosures. The pattern was different for waters at the limit of the Gironde plume and Atlantic oceanic waters (within salinity range 34.5-35.4). P would not be the single limiting nutrient of winter blooms and spring phytoplankton growth since low phosphate, and also low nitrate and silicate, availability were recorded and phosphate addition alone had no effect on phytoplankton biomass and production in bioassays. The early P limitation of winter blooms had consequences for the phytoplankton community structure in the Gironde plume waters (salinity < 34.5). Whereas major cells of these blooms were greater than 20 pm in size, phytoplankton growth in spring and autumn was dominated by 3-20 μm (e.g. 53% of Chl a in late April 1999) and < 3 mum cells (e.g. 29% of Chl a). The decreasing size of phytoplankton cells is emphasized by the severe competition between bacteria and algae for phosphate, since bacteria dominated phosphate uptake in spring (e.g. 87% in late April, 77% in late May). Bacteria tended to have greater affinity for phosphate and seemed also to be P limited at certain times in spring, according to results from phosphate enrichment bioassays in late May 1999. The alternative method for phytoplankton to obtain P would be the use of the dissolved organic phosphorus pool by alkaline phosphatase activity. According to the movement of P-33 after initial labelling of microbial populations and a subsequent cold chase, the major transfer of P occurred from the bacterial to the dissolved fraction. We hypothesize that algae obtain part of its dissolved organic phosphorus from bacteria-originated organic phosphorus compounds.