Outdoor experimental tanks were used to simulate phytoplankton blooms in natural seawater from the bay of Villefranche-sur-Mer (France), during March-April 1984. External temperature and light intensity were respectively monitored every 4 h and daily. Nutrient conditions were followed daily and maintained by a cycle of dilutions with fresh enriched seawater. Six different cycles of phytoplankton production were followed. They differed by the dilution rates used. Nutrient uptake. phytoplankton composition, biochemical composition and detailed lipid composition of the phytoplankton were followed throughout each growth cycle. Results showed a clear difference between the first cycle, which lasted 7 d , and subsequent ones which lasted only from 27 to 51 h In terms of species composition, the initial body of water con~prised a complex assemblage of species (flagellates, dinoflagellates, diatoms, microzooplankton) which were gradually replaced by a community of diatoms dominated by Skeletonema costatum, Nitzschia sp. and Chaetoceros sp. Total fatty acid changes reflected the progressive changes in species composition (inverse variation of C16 and C18 polyunsaturated fatty acids) and to a minor extent the nutrient regime. In terms of fatty acid composition of the different lipid classes the transition from a flagellate to a diatom population appears to b e the key factor influencing all categories in a similar way. Nutrient regime affected essentially the neutral lipid composition while changes in physiological state were reflected in the polar lipid composition (phospholipids, monogalactosyl diglycerides [MGDG], digalactosyl diglycerides). Probable light limitation during the fourth growth cycle may explain some of the changes in fatty acid composition at the MGDG level. In all cases, transient saturation-desaturation processes of the membrane lipids seems at the heart of the acclimation dynamics