Several experiments using a self-regulated system were conducted to define the factors likely to influence the uptake of paralytic shellfish poison (PSP) by oysters in the Penzé estuary (France, Brittany). Each 4-day experiment was carried out in a recirculated sea water system using 15 Pacific oysters (Crassostrea gigas) separated from each other and supplied with unfiltered natural seawater containing alternatively toxic (Alexandrium minutum) or non-toxic (Skeletonema costatum) algal diets. The food supply and exposure times to toxic diets were determined according to field studies of the upstream and downstream movement of patches containing A. minutum. The experimental parameters corresponded roughly to the hydrological conditions generally observed in June when tidal coefficients are lowest and blooms occur: (i) A. minutum concentrations in sea water of 200, 5000 and 10 000 cell ml−1; (ii) inorganic matter consisting of 5 and 15 mg L−1 of calcinated muddy sediments; and (iii) low and high tide salinities of 25 and 35‰, respectively. Significant experimental contamination (greater than the 80 µg STX equiv. 100 g-1 sanitary threshold) occurred after 4 days of exposure for the monospecific A. minutum diet (20-200 cell ml−1) and alternated A. minutum and S. costatum diets (5000 and 20 000 cell ml−1, respectively). Contamination levels were less than the sanitary threshold for alternated A. minutum/S. costatum diets of 200 and 20 000 cell ml−1, respectively, and for a monospecific A. minutum diet (1000−10 000 cell ml−1). In the last case, the accumulation rate was quite low, possibly because of inhibition of the filtration rate related to a lower biodeposit production rate and decreased feeding time activity. The addition of inorganic matter appeared to play a significant role in the observed increase of toxin uptake, whereas salinity was not a determining factor for toxin accumulation rates. These last observations were corroborated by statistical analysis and stepwise multiple linear regressions integrating all or some of the experimental parameters.