The chemical form of accumulated trace metal in prey is important in controlling the bioavailability of dietary metal to a predator. This study investigated the trophic transfer of radiolabelled Ag, Cd and Zn from the polychaete worm Nereis diversicolor to the decapod crustacean Palaemonetes varians. We used 2 populations of worms with different proportions of accumulated metals in different subcellular fractions as prey, and loaded the worms with radiolabelled metals either from sediment or from solution. Accumulated radiolabelled metals were fractionated into 5 components:metal-rich granules (MRG), cellular debris, organelles, metallothionein-like proteins (MTLP), and other (heat-sensitive) proteins (HSP). Assimilation efficiencies (AE) of the metals by P. varians were measured from the 4 categories of prey (i.e. 2 populations, radiolabelled from sediment or solution). There were significant differences for each metal between the AEs from the different prey categories, confirming that origin of prey and route of uptake of accumulated trace metal will cause intraspecific differences in subsequent metal assimilation. Correlations were sought between AEs and selected fractions or combinations of fractions of metals in the prey-MRG, Trophically Available Metal (TAM = MTLP + HSP + organelles) and total protein (MTLP + HSP). TAM explained 28% of the variance in AEs for Ag, but no consistent relationships emerged between AEs and TAM or total protein when the metals were considered separately. AEs did, however, show significant positive regressions with both TAM and total protein when the 3 metals were considered together, explaining only about 21% of the variance in each case. A significant negative relationship was observed between MRG and AE for all metals combined. The predator (P. varians) can assimilate dietary metal from a range of the fractions binding metals in the prey (N. diversicolor), with different assimilation efficiencies summated across these fractions. TAM and/or total protein may represent an approximate minimum for trophic availability but neither of these alone is a fully accurate predictor.