Previous data about silver in marine molluscs have shown a high variability of both accumulation and toxicity of this metal. A potential relationship between these factors has been examined. In a preliminary experiment, three species of bivalves ( Crassostrea gigas, Mytilus galloprovincialis, Chlamys varia) were exposed to silver concentrations varying from 1 to 1000 μg Agl −1 in order to determine parameters of lethal toxicity. Then, specimens of the same species were exposed to sub-lethal levels of silver through food (groups 2), water (groups 3) or both food and water (groups 4). In this main experiment, the level of silver was identical (20 μg Ag1 −1 in the culture medium of algae used as food and in the rearing medium of bivalves. Since the nominal levels of silver in water decreased during the experiment (Tables 1 and 2), seawater and contaminant were renewed every day. The species-related variations of mean lethal time (Table 3) revealed that the most resistant species was the oyster and the lesser one was the scallop. The resistance of oysters was mentioned previously by Martoja et al. (1988) who did not observe any cytological lesion in individuals exposed to silver in experimental conditions similar to those applied in the present study. The susceptibility of scallops has been confirmed since a moderate level of silver of 20 μg Ag l −1 was sufficient to disturb their attachment to the substratum (Table 4) as a consequence of a heavy accumulation of silver sulphide in glandular cells which secrete byssus threads (Martoja et al., 1990). The patterns of bioaccumulation in soft tissues were different in oysters—in which the maximal concentration was reached within 2 weeks (Fig. 1)—and in mussels—which this maximum was not reached after 4 weeks (Fig. 2). Exposure to silver via food induced a significant uptake of this metal in soft tissues of bivalves but it was low compared to uptake from seawater and to the contamination due to the double way of exposure (Table 5). However, the biomass of molluses per unit volume of culture medium varied from species to species. In order to compare strictly the specific ability to concentrate silver, we have determined the quantity of metal available (from food and/or water) per unit weight of filter-feeding organism over a definite period. Then we calculated the percentage of available silver retained by each species (Table 6). These results, as well as the literature data (Brooks and Rumsby, 1965; Folsom and Young, 1965; Preston et al., 1968; Bryan, 1973; Pouvreau and Amiard, 1974), show that Pectinidae and Ostreidae are able to retain very high levels of silver compared to Mytilidae. The species-related behaviour of silver may be interpreted from the point of view of ecotoxicological consequences: • -Species such as the oyster, with a low susceptibility and high ability to concentrate silver, could have an important role as the vector of contamination in food webs. • -Species such as the scallop, able to concentrate high levels of silver and very susceptible, could be eliminated from polluted ecosystens but, in these conditions, their role in trophic transfer will be limited. • -Species such as the mussel, which accumulate only low quantities of silver, will not have any role in food chain contamination and their intermediate susceptibility could have consequences only in highly polluted environments.