To test the hypothesis that treatment with antibiotics prevents low birth weight, pregnant women whose vaginal cultures contained Ureaplasma urealyticum or Mycoplasma hominis (or both) and who gave written informed consent were treated with one of the following: identical looking capsules containing 250 mg of either erythromycin estolate or stearate (active against U urealyticum), or 150 mg of clindamycin hydrochloride (active against M hominis), or placebo, four times daily for six weeks in a randomized double-blind study. Treatment with clindamycin had no effect. Treatment with erythromycin initiated during the second trimester had no effect on mean birth weight or on the frequency of low-birth-weight infants. In contrast, women whose treatment with erythromycin was initiated in the third trimester gave birth to infants with a heavier mean birth weight (3331 g) than infants born to placebo-treated women (3187 g) (P = .042). Similarly, in women whose erythromycin was begun during the third trimester, the birth rate of infants weighing 2500 g or less was 3%, whereas in women treated with placebo, the birth rate for low-birth-weight infants was 12% (P = .047). These data suggest that treatment with erythromycin during the third trimester prevents low birth weight in mycoplasma-colonized pregnant women. Whether the effect is due solely to the action of erythromycin on U urealyticum is uncertain.