The clinical significance, Gram stain reaction, and genus affiliation of Gardnerella vaginalis have been controversial since Gardner and Dukes described the organism as the cause of "nonspecific vaginitis," a common disease of women which is now called bacterial vaginosis. The organism was named G. vaginalis when taxonomic studies showed that it was unrelated to bacteria in various genera including Haemophilus and Corynebacterium. Electron microscopy and chemical analyses have elucidated the organism's gram-variable reaction. Controversy over the etiology of bacterial vaginosis was largely resolved by (i) studies using improved media and methods for the isolation and identification of bacteria in vaginal fluids and (ii) standardization of criteria for clinical and laboratory diagnosis. Besides G. vaginalis, Mobiluncus spp., Mycoplasma hominis, and certain obligate anaerobes are now acknowledged as participants in bacterial vaginosis. The finding that G. vaginalis, Mobiluncus spp., and M. hominis inhabit the rectum indicates a potential source of autoinfection in addition to sexual transmission. Extravaginal infections with G. vaginalis are increasingly recognized, especially when the toxic anticoagulant polyanetholesulfonate is omitted from blood cultures and when urine cultures are incubated anaerobically for 48 h. The finding that mares harbor G. vaginalis suggests that an equine model can be developed for studies of Gardnerella pathogenesis.