Potamocorbula amurensis was assessed as a biosentinel species in San Francisco Bay, California, USA. Uptake of metal in both the laboratory and field showed that P. amurensis was sufficiently responsive to Ag, Cd, Cr, Ni and V to detect environmental differences in exposure. It was less suitable as an indicator of Cu and Zn contamination. Concentration factors for P. amurensis were: Ag, 386000; Cd, 50200; Cr, 36600; Cu, 12200; Ni, 5200; and Zn, 115500. Samples were collected from 6 stations throughout the bay at near-monthly intervals from January 1991 to March 1992. Variability within a collection was influenced by gut content and animal size. Other sources of variability were time [coefficient of variation (CV) = 10 to 21%], small-scale spatial variability (within 3 km, CV = 10 to 25%), and large-scale spatial variability (CV = 3.3 to 12.4%). Depuration for 48 h was necessary to mitigate bias from gut content. Precision was improved by analyzing large numbers of individuals (60 to 120) separated into several (5 to 14) composites at each collection and by determining, from regression, the mean and variance for samples with significant correlations between metal concentration and shell length. Repeated monthly sampling increased the accuracy of long-term site characterizations. Temporal variability was small because of drought. The grand means of the concentrations of Ag, Cd, Cr, Ni, and V in the tissues of P. amurensis at each station for the 15 mo period revealed persistent contamination from industrialized Suisun Bay to the mouth of San Francisco Bay. Demonstration of responsiveness, precision and accuracy should be a prerequisite for the optimal use of biosentinels.