The presence of detectable levels of polychlorinated biphenyls (PCBs) in commercially produced fish feed has raised a concern about the degree of biomagnification of these contaminants in hatchery-reared trout. Our objectives were to (1) define the relationship between concentrations of PCBs in fish feed and in fish tissue and (2) estimate the relative contributions of feed and hatchery supply water to PCB concentrations in fish. We conducted a 6-month feeding trial with fingerling rainbow trout Oncorhynchus mykiss fed commercial diets with four concentrations of PCBs: a low-PCB diet (69 ng/g); a typical commercial diet (126 ng/g); and the typical diet spiked with PCBs at two levels (220 and 280 ng/g). The concentrations of PCBs in fillets after 1 month were commensurate with those in the feeds and remained relatively stable for the next 5 months; mean PCB concentrations in fillets ranged from 54 to 94 ng/g. Low levels of PCBs were detected in the hatchery supply water. We used the concentrations of PCBs in the feeds, absorption rates of PCBs, and two different rates of PCB depuration to estimate the potential uptake of PCBs from supply water. When we used a low depuration rate (half-life = 219 d), the computed body burdens of PCBs could be entirely attributed to the feeds. When a high depuration rate (half-life = 66 d) was used, some uptake of PCBs from the supply water was likely, but most of the total body burden originated from the feeds. We concluded that rainbow trout fed a diet with 126 ng/g PCBs would have a PCB concentration of about 60 ng/g in their fillets, which is high enough to warrant issuance of a consumption advisory (no more than one meal of fish per week) under a protocol adopted by some Great Lakes states.