Hepatocytes were isolated from wild-type and connexin32-deficient (Cx32-deficient) mice. Pairs of cells were chosen to study the electrical properties of gap junction channels using the dual voltage-clamp method. The total gap junction currents revealed that Cx32-deficient hepatocytes express one type of connexin (Cx26) and wild-type hepatocytes express two types of connexins (Cx26 and Cx32). The unitary gap junction currents suggest that Cx32-deficient cells have homotypic channels (Cx26-Cx26) while wild-type cells form homotypic (Cx26-Cx26, Cx32-Cx32) and heterotypic channels (Cx26-Cx32). Homotypic channels exhibited a main conductance and a residual conductance, both virtually insensitive to gap junction voltage (Vj) (Cx32-Cx32: gammaj,main=31 pS, gammaj,residual=9 pS; Cx26-Cx26: gammaj,main=102 pS, gammaj,residual=17 pS). Residual states were regularly seen in Cx32-Cx32 channels, but rarely in Cx26-Cx26 channels. Heterotypic channels showed a main conductance and a residual conductance. The former was sensitive to Vj (average gammaj,main=52 pS). The electrophysiological data suggest that Cx32 hemichannels are more abundant than Cx26 hemichannels in prenatal (ratio 4:1) and adult wild-type hepatocytes (ratio 23:1) and that the total number of gap junction channels is larger in prenatal cells than in adult cells. The diversity of the relationship gj, ss/gj,inst=f(Vj) (gj,ss: gap junction conductance at steady state; gj,inst: instantaneous gap junction conductance; Vj: transjunctional voltage) seen in wild-type cells suggests that the ratio Cx26/Cx32 hemichannels is variable among hepatocytes. A comparison of total and unitary conductances implies that Cx26 hemichannels are down-regulated in Cx32-deficient cells and that docking between Cx26 and Cx32 hemichannels occurs randomly. While the gap junction currents are compatible with homotypic and heterotypic channels, the presence of heteromeric channels cannot be excluded.