A total of 952 consecutive vaginal swabs were obtained from patients who attended obstetric or gynecologic clinics affiliated with the Children's Hospital of Buffalo, New York. Swabs were cultured comparatively on 5% sheep blood agar (BA), selective sheep blood agar containing 1.25 mu g/ml trimethoprim-23.75 mu g/ml sulfamethoxazole (SXT), and Lim broth (Todd-Hewitt broth containing 1% yeast extract, 1O mu g/ ml colistin, and 15 mu g/ml nalidixic acid). A total of 168 swabs (18%) were positive (by at least one method) for group B Streptococcus (GBS). The overall agreement among the three techniques was 90% (858 of 952); 94 specimens (10%, 94 of 952) had discrepant results, and 74 of these (44%, 74 of 168) were positive, only by Lirn as opposed to two (1%) and 0 by BA and SXT, respectively. There were only two (2%, 2 of 168) false negative for Lim as compared with 82 (49%) for BA and 86 (51%) for SXT. Thus, the sensitivity of css detection by BA, SXT, and Lim is 51%, 49%, and 99%, respectively. These data suggest that the use of Lim broth increases the recovery rate of GBS by 48% after 48 h while the use of the SXT plate reduced the recovery rate by 2% as compared with the conventional BA plate.