The toxicity of PCB-156 (2,3,3',4,4',5-hexachlorobiphenyl) was investigated in rats following subchronic dietary exposure. Groups of 10 male and female Sprague-Dawley rats were administered PCB-156 in the diet at 0, 0.01, 0.1, 1 or 10 ppm for 90 days. Dose-dependent increases were detected for the liver, lung and kidney weights, as well as for the liver EROD, PROD and UDPGT enzyme activities and liver uroporphyrin concentration. Dose-dependent decreases were observed in final body weight, body weight gain, and thymus weight. Apolar retinoid concentrations were decreased in the liver and lungs and increased in the kidneys. Histopathological examination of the liver, thyroid, and thymus showed mild to moderate dose-related changes. A LOAEL of 0.01 ppm was established, based on reduced apolar liver retinoid concentration. Benchmark dose-modelling corroborated the sensitivity of liver retinoid endpoints. The lower confidence limits (BMDL) for a 5% decrease in apolar liver retinoid concentrations were 0.0009 and 0.0007 ppm, respectively, in males and females, corresponding to a daily dose of 0.06 μg PCB-156 per kg body weight. Organizing dose-response data for the individual hepatic endpoints along the PCB-156 dosing scale revealed a sequence of events compatible with a causal link between depletion of apolar retinoids and the other liver biochemistry and pathology findings. Taken together, data suggest that the retinoid endpoints should be further evaluated for a causal relationship to PCB-induced liver toxicity and that retinoid system endpoints are identified and characterized to support health risk assessment in the emerging research fields of endocrine disruption and mixture toxicology.