This cohort study of 182 women attending a sexually transmitted disease clinic evaluated the hypothesis that women colonized by lactobacilli have decreased acquisition of vaginal infections. During a 2-year follow-up, 50 women acquired bacterial vaginosis (BV), 25 acquired symptomatic vulvovaginal candidiasis (VVC), and 7 acquired vaginal trichomoniasis. By multivariate analysis, utilizing Cox proportional hazards modeling with time-dependent covariates, acquisition of BV was independently associated with lack of vaginal H2O2-producing lactobacilli (hazard ratio [HR] = 4.0, P < .001) or presence of only non-H2O2-producing lactobacilli (HR = 2.2, P = .02). Acquisition of BV was associated with having a new sex partner (HR = 2.5, P = .004) and with douching for hygiene (HR = 2.1, P = .05). Absence of lactobacilli did not increase acquisition of VVC. Trichomoniasis was associated only with having a new sex partner (HR = 4.7, P = .05). These results support the hypothesis that H2O2-producing vaginal lactobacilli protect against acquisition of BV but do not protect against VVC or vaginal trichomoniasis.