We have calculated tunneling conductance in metal–narrow-gap-semiconductor (NGS)–metal tunnel junctions. Flietner's two-band model is used to describe the dispersion relation within the energy gap in an isotropic NGS with different conduction- and valence-band edge effective masses. The results are compared with the tunneling conductance calculated by Kane's two-band model, which has been commonly used to describe the tunneling characteristics through the energy gap in semiconductors. These results propose that the tunneling conductance in the tunnel junctions in which a narrow gap semiconductor of largely different conduction- and valence-band effective masses is used as a tunneling barrier can exhibit quite a different behavior, especially in the region of the midgap, from the tunneling conductance described by Kane's two-band model.