One hundred fifty-six women with preterm labor between 24 and 34 weeks' gestation were randomized to receive either intravenous magnesium sulfate or no tocolytic therapy. Magnesuim sulfate infusions of up to 3 gm/hr were used in 76 pregnancies and resulted in a mean serum magnesium concentration of 5.5 +/- 1.4 mEq/L (mean +/- SEM). Compared with 80 control pregnancies, magnesium sulfate tocolysis had no significant effect on duration of gestation, birth weight, neonatal morbidity, and perinatal mortality. We conclude that clinically safe infusions of magnesium sulfate are ineffective when used to prevent preterm birth.