To evaluate the usefulness of quantitative cultures of Gardnerella vaginalis in the laboratory determination of nonspecific vaginitis, the actual and relative numbers of G. vaginalis in genital cultures of a general patient population were assessed semiquantitatively, and the laboratory results were then correlated with the clinical findings. Of the 1,585 women studied, 417 (26.3%) yielded G. vaginalis in culture. Of these, only 113 (27.1%) were found to have symptoms and signs consistent with nonspecific vaginitis. G. vaginalis was obtained in pure or predominant growth from 87 of 100 consecutive cases with nonspecific vaginitis and 32 of 100 consecutive cases without the symptoms or signs of vaginitis (P less than 0.001). Hence, the positive predictive value of isolation of G. vaginalis in pure and predominant growths was determined to be 73% (87 of 119). Conversely, G. vaginalis was isolated in mixed or light growth significantly more often from asymptomatic women than from symptomatic patients, i.e., 68 versus 13 cases. Therefore, the negative predictive value of isolation of G. vaginalis in mixed and light growths was found to be 84% (68 of 81). Quantitation of the relative amount of G. vaginalis growth had higher predictive values as compared with the assessment of G. vaginalis growth alone. We conclude that quantitative culture of G. vaginalis is essential to obtain maximum reliability of culture results in the laboratory determination of nonspecific vaginitis. Although quantitated cultures of G. vaginalis have high predictive values, laboratory results must be interpreted in conjunction with the clinical findings.