This study examined the relationships between population characteristics and the expression of physiological biomarkers of stress in an intertidal clam population under pollution at sites differing in thermal history and coastline distance. The clam population metrics were age distribution, growth, condition factor, distance of the clam beds from the shore, and gonad development. Physiological biomarkers comprised biomarkers of defence such as superoxide dismutase, labile IIb metals in tissues, redox status of metallothioneins and glutathione S-transferase, of tissue damage such as lipid peroxidation and DNA strand breaks, of reproduction as determined by vitellogenin-like proteins and gonadosomatic index and immunocompetence such as phagocytosis and hemocyte viability. Age-related pigments were also examined to compare the physiological age of the clams with their chronological age. The results showed that all the above biomarkers were significantly affected at one of the two polluted sites at least. Distance from the shore was significantly correlated with most (81%) of the biomarkers examined. Clams collected at one polluted site were physiologically older than clams from the corresponding reference site. Canonical and adaptive regression (artificial neural networks) analyses found that the biomarkers measured in this study were able to predict the ecologically relevant endpoints. Biomarkers implicated in defense mechanisms, tissue damage and age-related pigments were most closely related to the clam population characteristics. Sensitivity analysis of the learning algorithm found that the following physiological and biochemical markers were the most predictive, in decreasing order, of clam population characteristics: glutathione S-transferase, phagocytosis, age pigments, lipid peroxidation in the gills, labile IIb metals and total MT levels. These biomarkers were affected by the distance of the clam beds from the shore, site quality (pollution) and reproduction activity.