Human connexins 26 and 30 were expressed either through the bicistronic pIRES-EGFP expression vector or as EYFP-tagged chimeras. When transiently transfected in communication-incompetent HeLa cells, hCx26-pIRES transfectants were permeable to dyes up to 622 Da, but were significantly less permeable to 759 Da molecules. Under the same conditions, permeability of hCx26-EYFP fusion products was comparable to that of hCx26-pIRES, but with significant increase in diffusion at 759 Da, possibly as a consequence of having selected large fluorescent junctional plaques. Dye transfer was limited to 457 Da in hCx30-EYFP transfectants. When reconstructed from confocal serial sections, fluorescent plaques formed by hCx26-EYFP and hCx30-EYFP appeared irregular, often with long protrusions or deep invagination. Similar plaques were observed following immunostaining both in cells transfected with hCx26-pIRES and in HeLa cells stably transfected with mouse Cx26. Tissue conductance (Tg(j)) displayed significantly smaller values (28.8+/-1.8 nS) for stably transfected mCx26 than transiently transfected hCx26 (43.5+/-3.3 nS). These differences reflected in distinct functional dependence of normalized junctional conductance (G(j)) on transjunctional voltage (V(j)). The half-activation voltage for G(j) was close to +/-95 and +/-58 mV in mCx26 and hCx26, respectively. The corresponding parameters for hCx30 transfectants were Tg(j)= 45.2 +/- 3.5 nS and V(0)= +/- 34 mV. These results highlight unexpected differences between mCx26 and hCx26 in this expression system, reinforce the concept that channel permeability may be related to Cx level expression, and indicate that fusion of hCx30 to GFP colour mutants produces channels that are suitable for permeability and gating studies.