Risk assessment of metals in the environment is performed mainly with toxicity evaluations on single metals, which is largely inadequate since these substances occur in mixtures. The development of models predicting combined toxic effects on the basis of the concentration-response relationships of individual compounds has emerged as an answer. In the present study, metal effects on post-exposure anorexia (the concept of FdC(50)--concentration causing 50% of feeding inhibition--is implemented) in Echinogammarus marinus, a widely distributed gammarid amphipod, were assessed and compared with modelled ones obtained through the application of the concentration addition (CA) model, which represents a reasonable worst-case scenario for the risk assessment of metal mixtures. Data were validated using in situ experiments performed along a latitudinal gradient (Iceland, Scotland and Portugal) aiming at establishing a geographic profile of autochthonous population susceptibilities to metals. For all of the metals studied concentrations in the water column at exposure sites were in good agreement with feeding inhibition levels. Models gave low to relatively high percentage agreement between predictions and experimental data. Boreal populations demonstrated higher susceptibility to single metals, but not to mixture exposures. Meridional populations denoted lower susceptibilities with higher FdC(50).