This paper reports the numerical analysis of 100 strains of Candida albicans using 96 physiochemical characters. Principal co-ordinate analysis and the modified multidimensional group analysis (MGA) method were used for the numerical analysis. MGA detected optimal solutions for five, 11, 15 and 17 groups. A single large swarm represented the majority of strains for the five- and 11-group solutions. Of the remaining two solutions only the 17-group solution was analysed further as the 15-group solution was close to the 17-group solution and contained less information. Of the 17 groups within the data, 10 “groups” contained only a single member, whilst the other groups contained 32, 24, 14, 13, three, two and two members each. Characters that tended to distinguish between the four larger groups included colonial fringe state, valerate assimilation, and resistance to borate, cetrimide, benzalkonium chloride, chlorhexidine, arsenate and salt. These larger groups appeared to form reasonably distinct clusters when tested by Sneath's overlap statistics. However, as the squared Euclidean distances between these group centroids were small, it would appear that most strains of C. albicans form a single swarm. It is suggested that these groups may represent “preferred phenotypes” for C. albicans, between which individual strains may switch.