Background/introduction In a multicenter clinical trial funded by BD, we observed less accurate clinician diagnosis of bacterial vaginosis (BV) based on clinical observations when Trichomonas vaginalis (TV) and/or Candida spp. were also detected by the trial Reference Methods than when only BV was detected. Aim(s)/objectives To determine the sensitivity of each criterion and of the overall Amsel’s criteria (3/4 criteria met), the results of the Amsel’s corresponding to the sub-population of specimens that gave a Nugent score of 7–10 were analysed. Methods Following informed consent, women with symptoms of vaginitis/vaginosis were included in the trial. The four Amsel’s criteria and the Nugent score were performed. Evaluation for trichomoniasis by wet mount and culture (InPouch™ TV, Biomed) were performed. Candida colonies were isolated (BBL™ Sabouraud Dextrose Agar, Emmons and BBL™ CHROMAgar™ Candida plate, BD) and identified by ITS-2 bi-directional sequencing (Accugenix®). Results In total, 269/497 (54.1%) specimens gave a Nugent score of 7–10. Amongst them, TV and/or Candida spp. were found in 100 specimens (37.2%). The sensitivity of clue cells, amine test, vaginal pH, BV vaginal discharge, and overall Amsel’s criteria in absence of TV and/or Candida spp. was 86.3%, 82.7%, 91.1%, 71.0%, and 84.6% respectively. In presence of TV and/or Candida spp., the sensitivity was 63.6%, 64.0%, 75.0%, 42.0%, and 60.0% respectively (p values ≤ 0.0009 for all comparisons). Discussion/conclusion The sensitivity of the Amsel’s criteria in women with BV decreases when TV and/or Candida spp. are present. The BV vaginal discharge is the least sensitive criterion.