Sixteen hundred and thirty seven isolates of Neisseria gonorrhoeae isolated over a two year period were serotyped using three panels of monoclonal antibodies. The isolates comprised 687 serogroup WI strains and 950 serogroup WII/III strains. The antibodies used in the panels were those developed by Pharmacia (Ph) and Genetic Systems (GS). The GS antibodies were used as two separate panels; the American (GS-A) panel which designates serovars by a simple numerical nomenclature and the Swedish (GS-S) panel which designates serovars by a more detailed descriptive nomenclature. Using only a single panel the Ph panel gave the greatest discrimination yielding 42 serovars compared with 28 serovars with the GS-S panel and 23 serovars with the GS-A panel. Using two panels in combination, the GS-A/Ph panel combination gave the greatest discrimination with 81 serovar combinations while a combination of the GS-S and Ph panels yielded 77 serovar combinations. A compilation of antibodies from all three panels yielded 90 serovar combinations. As the combination of GS-A and Ph panels gave a greater degree of discrimination than the combination of the GS-S and Ph panels we advocate that first-stage serotyping should be performed with the more widely used GS-A panel while second-stage serotyping should be performed with the Ph panel. We propose that serovar combinations should be reported using a dual nomenclature e.g. IA-1/Arost, IB-1/Bropt, IB-1/Bropyt etc. The use of such a dual system would allow "core comparison" between all centres while maintaining a degree of flexibility regarding the extent of discrimination required.