SUMMARY Seven hundred and ninety three women were investigated, aged between 16 and 78 years, to evaluate the prevalence of bacterial vaginosis (BV) and some associated micro-organisms, and to discuss the significance of laboratory indices correlated to this pathology. BV was diagnosed on the basis of four distinct criteria: a positive result of the test for amines with 10% KOH (odour-test), the presence of clue cells on fresh microscopic examination, a pH > 4 5 and direct Gram stain positive (the presence of more than 40 Gram negative or Gram variable coccobacilli per microscopic field by 1000 magnifications under oil immersion). The total prevalence of BV was 20-5% (163); similar percentages were found in both fertile and pregnant women, whereas a lower percentage (12 7%) was found in menopausal women. Gardnerella vaginalis was present in 235 (29-6%) of the 793 women, in 144 (88-3%) of the 163 with BV and in 91 (14-4%) of the 630 women without BV. Mobiluncus species was present in 8-2% (65) ofthe total population, in 38-6% (63) of the women with BV and only in two (0 3%) of the women without BV. In the women with BV lower percentages were found for Trichomonas vaginalis, yeasts, Chlamydia trachomatis and Neisseria gonorrhoeae. The absence of a definite relationship between BV and cultural isolation of G vaginalis is confirmed whereas the role played by Mobiluncus spp still has to be clarified. It is concluded that it is not necessary to screen with all four laboratory indices. Two positive indices from a panel of three (excluding pH > 4 5 and direct Gram stain positive in the same panel) allows the correct diagnosis of BV in almost all cases.Bacterial vaginosis (BV), a syndrome of probably mixed bacterial aetiology without signs of inflammation of the vaginal mucosa (also known as nonspecific vaginitis), has been studied over a long period to establish the aetiological role of a particular microorganism, Gardnerella vaginalis. Gardner and Dukes'2 were the first to establish a connection between the micro-organism and the disorder over 30 years ago.Although G vaginalis has been biochemically identified in eight biotypes,3 there are still no definite findings and no clear pathogenicity for any of these in relation to BV. The association, however, remains between the syndrome and the micro-organism (once called Haemophilus vaginalis or Corynebacterium vaginalis), and is confirmed in a number of studies.'