The bioaccumulation and toxicokinetics of 42 polychlorinated biphenyls (PCBs) was determined in male American kestrels exposed to an Aroclor-contaminated diet for 120 d followed by a 348-d depuration period. The birds were housed under ambient outdoor temperatures to permit normal fluctuations in body weight during the study. Whole body PCB clearance, plasma/fat distribution coefficients, and plasma PCB clearance constants were determined for individual PCBs to calibrate a two-compartment rate constant model in order to describe PCB elimination in the birds. Plasma/fat partition coefficients (K(PF)) averaged 0.0060 +/- 0.0001 for all congeners of study, were not dependent on chemical hydrophobicity, and did not change in summer versus winter sacrificed animals. Plasma clearance constants (k'pc) for PCBs were observed to be dependent on both chlorine substitution patterns and congener hydrophobicity. Polychlorinated biphenyl congeners categorized as readily cleared congeners contained vicinal meta-para hydrogen substituents on at least one phenyl ring, while slowly cleared congeners were chlorine hindered at these positions. A general equation was derived to predict plasma clearance constants for all tri- to octachlorobiphenyls based on the presence of an open meta-para site on one of the phenyl rings and from the n-octanol-water partition coefficient of the chemical. The equation was validated by comparing predicted versus measured relative biomagnification factors of PCBs determined in birds at the end of the dosing period. The two-compartment model calibrated for PCB elimination in American kestrels may be used to describe PCB toxicokinetics in wild birds provided that seasonal fluctuations in the fat content of the modeled population is known.