The G protein sequences of fourteen animal rhabdoviruses, representing all four recognized genera (Vesiculovirus, Lyssavirus, Ephemerovirus and Novirhabdovirus) and the ungrouped sigma virus, were aligned using CLUSTAL W and adjusted to account for obvious sequence similarities not detected by the algorithm. Analysis of the alignment indicated remarkable preservation of G protein structural features including cysteine residues, antigenic sites and significant elements of secondary structure (α-helices, β-strands and loops). Twelve highly conserved cysteine residues were assigned numbers (C I to C XII ) according to their location in the alignment. Other cysteine residues were assigned numbers (C 0 to C XIIe ) according to their position relative to the conserved cysteines. The pattern of conservation of cysteine residues and the structural characteristics of identified discontinuous antigenic sites were used to deduce a model for G protein structure. Six absolutely conserved cysteines are predicted to associate in three disulphide bridges (C I -C XII ; C VIII -C XI ; C IX -C X ) that form the core of the G protein structure and define the common discontinuous antigenic site. The associations of six other highly conserved cysteines (C II -C IV ; C III -C V ; C VI -C VII ) are predicted by the absence of a specific pair in all viruses within a genus. Of the other cysteines, one pair occurs only in ephemeroviruses and novirhabdoviruses (C 0 -C XIIa ) ; two pairs occur only in ephemeroviruses (C Ib -C VIIIa ; C XIIb -C XIIe ) ; and two pairs occur only in lyssaviruses (C Ia -C VIIIb ; C XIIc -C XIId ). The structures predicted by the model account for the preservation of conformational antigenic sites, accommodate genus-specific variations, and are generally consistent with previous observations of G protein structure.