To obtain efficient induction by phenobarbital (PB) of cytochrome P450 (CYP) genes in primary hepatocyte culture, previous studies have demonstrated the importance of culturing hepatocytes on a substratum derived from extracellular matrix (ECM or Matrigel), or in highly enriched medium formulations such as Chee's. In the present study, we have reexamined a variety of hepatocyte culture conditions and their relative abilities in preserving the PB-induction response within the CYP2B and 3A gene subfamilies. A modified culture system was developed that combines a highly effective medium formulation in conjunction with dilute concentrations of a Matrigel overlay. Specifically, hepatocytes were attached to a Collagen I substratum and overlaid with either daily (50 micrograms/ml medium) or single (233 micrograms/ml) additions of Matrigel. The PB-induction response obtained in vitro closely resembled that occurring in vivo. Induction in culture by 1 mM PB of CYP2B1, 2B2, and 3A1 mRNA levels was highly dependent on a variety of factors, including medium formulation and 0.1 microM dexamethasone addition. The response to dexamethasone addition on this gene battery varied in a medium-specific manner. Cells maintained a higher level of PB-induction response when maintained in Williams' E medium than with Chee's, Waymouth's, or Ultraculture media. The kinetics of PB induction also were more rapid in cells cultured in Williams' E medium. PB exposures in Chee's medium resulted in elevation of two CYP2B-immunoreactive proteins as detected by Western blot analysis, together with increased rates of O-dealkylation of benzyloxy- and pentoxyresorufin. However, Chee's formulation produced an abnormal PB-induced expression of CYP1A1, as determined by mRNA analysis, high rates of O-dealkylation of 7-ethoxyresorufin, and inhibition of enzymatic activity by 1 microM alpha-naphthoflavone. This paradoxical expression of CYP1A1 was not observed in PB-treated cultures grown in Williams' E medium. Thus, these studies demonstrated that the use of a physiologically balanced medium, i.e., Williams' E formulation, together with an overlay of ECM, preserves PB-induction responsiveness closely resembling that observed in vivo, and should better facilitate mechanistic investigations into the molecular nature of PB induction.