Thermostratification and seasonal Light increase are generally considered the first causes of phytoplankton spring blooms in temperate waters. The objective of this study is to confirm the existence of winter phytoplankton blooms, responsible for the early exhaustion of phosphate, within the Gironde plume waters (southeast Bay of Biscay), and to understand what may initiate them so early. Two cruises, BIOMET 2 and BIOMET 3, were carried out respectively in early (8 to 21 January) and late winter 1998 (25 February to 11 March). An increase of phytoplankton biomass (chl a) between the 2 cruises and non-conservative nutrients observed in late winter confirm an early phytoplankton growth both in the Gironde plume and adjacent oceanic waters. Among factors that possibly initiate these blooms, light availability seems to be the best candidate. First, the status of available Light for phytoplankton changed dramatically between the 2 cruises when comparing the instantaneous depth-averaged irradiance (E-m) received by phytoplankton with the light saturation parameter E-k determined from P versus E curves. Light was limiting for phytoplankton growth in early winter according to systematically lower values of E-m than E-k in the daytime. However, light was not limiting during a large part of the day in late winter since E-m was above E-k during 40% of the daylight. The estimated critical depth from the Nelson and Smith (1991; Limnol Oceanogr 36:1650-1661) reformulation of the Sverdrup (1953; J Cons Perm Int Explor Mer 18:287-295) equation was much shallower (4 to 13 m) than the mixed layer depth (15 to 50 m) in early winter in the whole Gironde plume up to oceanic waters. On the contrary, it was much deeper (35 m) than the mixed layer depth (9 m) in late winter for the intermediate plume (salinity range 33 to 35) and similar for oceanic waters (S > 35.5) with a bottom depth of 60 to 70 m. Therefore, according to the Sverdrup model, available light is the triggering factor of winter phytoplankton blooms in the Gironde plume and adjacent oceanic waters. Moreover, Riley's (1957; Limnol Oceanogr 2:252-270) empirical critical irradiance of 20.9 W m(-2) could be applied in these waters to predict the onset of blooms since it was reached in late winter but not in early winter. The establishment of halostratification and/or decrease of the light vertical attenuation coefficient prevail in the initiation process of winter phytoplankton blooms.