Bacterial vaginosis (BV) is characterised by a depletion of lactobacilli in favour of an overgrowth of anaerobic bacteria. It is associated with increased risk for urogenital infections and abortion. In this study we assessed the effect of a yoghurt drink containing Lactobacillus strains on BV. The strains had been isolated from healthy pregnant women and selected for acidification capacity, production of H₂O₂, glycogen utilisation, bile salt tolerance and inhibition of pathogens. Using Amsel criteria BV was diagnosed in 36 women aged ≥18 years with stable menstrual cycle or menopause. They were treated with oral metronidazole for 7 days (2×500 mg/d). Starting with the treatment, women consumed twice daily either verum or placebo during 4 weeks. Verum was 125 g yoghurt containing (besides Lactobacillus delbrueckii ssp. bulgaricus and Streptococcus thermophilus) living strains Lactobacillus crispatus LbV 88 (DSM 22566), Lactobacillus gasseri LbV 150N (DSM 22583), Lactobacillus jensenii LbV 116 (DSM 22567) and Lactobacillus rhamnosus LbV96 (DSM 22560), each 1×107 cfu/ml; placebo was 125 g chemically acidified milk. After 4 weeks of intervention 0 of 17 had BV in the verum group versus 6 of 17 in the s.a. control (0.018 in Fisher Exact test). Amsel score decreased during the intervention period by 4.0 (median) (4.0; 3.0) (25^th; 75^th percentile) in the verum group compared to 2.0 (4.0; 0.0) in the control group (P=0.038 in Mann-Whitney test). Discharge and odour (Amsel criteria 2+3) also decreased by 2.0 (2.0; 1.0) in the verum compared to 1.0 (2.0; 0.0) in the control group (P=0.01) and differed after 4 weeks intervention between the groups 0.0 (0.0; 0.0) versus 1.0 (0.0; 2.0) (P=0.001). Nugent score decreased during the intervention period by 5.5 (7.0;2.3) in the verum compared to 3.0 (6.0;0.5) in the control group (P=0.158). Additional intake of yoghurt containing these probiotic strains improved the recovery rate and symptoms of BV and tended to improve the vaginal microbial pattern.