We conducted a retrospective study of outcome after cardiopulmonary resuscitation (CPR) in babies of very low birth weight. Of 158 such babies (birth weight, less than 1500 g) admitted to a neonatal intensive care unit in 1985, 49 (31 percent) underwent CPR. Low birth weight, low Apgar scores, birth asphyxia, pulmonary interstitial emphysema, hyaline membrane disease, and severe intraventricular hemorrhage were associated with the need for CPR. None of the 38 babies who received CPR in the first three days of life survived. Four of the 11 babies who received CPR after the first 72 hours survived. Three of the four survivors had residual neurologic deficits. Survival rates after CPR in infants of very low birth weight are lower than those in older children or adults. CPR may therefore be considered a nonvalidated therapy in this population. If the results of our study are confirmed, CPR should not be instituted automatically in very-low-birth-weight babies as though it were a validated treatment. Instead, it should be administered upon parents' advance informed consent to experimental treatment.