The use of gonococcal serovars in studying the epidemiology of Neisseria gonorrhoeae is well established. Most studies assume that the isolated serovar remains stable in vivo indefinitely. This study was designed to observe the correlation between serovars isolated in patients naming each other as sexual contacts. The overall rate of discordant transmission episodes was 12% (26/220). There were however significantly more discordant transmission episodes for partners of patients infected with IB serovars than IA serovars: 19% (23/124) versus 3% (3/96) -p < 0.01. The overall prevalence of serogroup IB isolates although significantly higher at 53% (p < 0.01) was thought insufficient to account for the correlation between discordant pairs and serogroup IB infection. Reasons considered for the discrepancies in contact pairs included: problems of partner reporting involving inaccurate or incomplete information; technical problems with reagents; culture induced phenotypic variation in antigenic expression or differences in the in vitro recognition of epitopes; and antigenic differences resulting from genetic mutation within the Protein I gene. It was concluded that carefully planned and controlled prospective studies involving duplicate parallel testing of isolates from patients and their well documented partners are needed to assess the extent to which these various factors contribute to discordancies in serovars isolated from contact pairs. Serotyping should be combined with other methods such as auxotyping for detailed microepidemiological studies involving partner notification.