OBJECTIVE: Bacterial vaginosis causes symptomatic vaginal discharge and has been associated with pre-term birth and with the acquisition of human immunodeficiency virus. Half of all women with bacterial vaginosis are free of symptoms, and treatment of these women is controversial. The objective of this study was to determine the extent of poor symptom recognition in this group of women. STUDY DESIGN: Seventy-five women attending a sexually transmitted disease clinic who had asymptomatic bacterial vaginosis were entered into a randomized, double-blind, placebo-controlled trial comparing metronidazole gel with placebo. Subjects' perceptions about changes in vaginal discharge and odor were determined, and treatment and placebo groups were compared by means of standard statistical analysis. RESULTS: When subjects were stratified by treatment group, there were no differences in their retrospective assessments of vaginal discharge and odor. A subset of women who had normalization of clinical parameters or of Gram stain scores did admit retrospectively to improvement; however, the difference between this group and the group without normalization was not statistically significant. Twenty-one percent of treated women subsequently had vaginal candidiasis. CONCLUSIONS: A greater percentage of women with resolution of bacterial vaginosis did retrospectively notice improvement in vaginal discharge and odor in comparison with those women without resolution; however, this was not statistically significant. These findings do not support routine treatment of women with asymptomatic bacterial vaginosis.