OBJECTIVE To determine mean arterial pressure values during the first 24 hours for "stable" and "unstable" extremely low birth weight (ELBW) infants and to ascertain its association with perinatal factors. BACKGROUND In ELBW infants, hypotension is diagnosed by nonspecific clinical signs together with reference arterial pressure values extrapolated from regression models or from scarce actual observations. METHODS Retrospective cohort study. METHODS 101 ELBW (< or = 600 g) infants born in our medical center (1989-2000). Considered stable were 36 infants with umbilical cord hemoglobin > or = 14 g/dl who, although mechanically ventilated, had normal acid-base balance, no patent ductus arteriosus, had not received indomethacin, steroids, muscle relaxants, narcotics, were never treated for hypotension and survived at least 7 days. The remaining 65 infants constituted the unstable group. Arterial pressures were determined by oscillometry (OBP) and direct transducer readings through an umbilical line (MAP). All admission and 10% of the readings were by OBP; the remaining 1877 measurements were by MAP. RESULTS Stable and unstable infants were similar in birth weight, demographics, history of chorioamnionitis, antepartum steroids, low Apgar scores, administration of epinephrine during resuscitation, and sepsis. Stable infants were different from unstable in gestational age (27+/-2 vs 25+/-2 weeks' gestational age [w GA]), history of preterm labor, preeclampsia, and neonatal mortality (22 vs 68%). Admission OBP (30+/-7 vs 29+/-10 Torr) were similar and 1-hour MAP were different (30+/-6 vs 27+/-7 Torr) between both groups. MAP for stable infants was higher throughout the 24 hours. Greater differences were noted between 3 and 6 hours when 34 of 65 unstable infants were treated for hypotension. Mean MAP and 10 percentile values for stable infants at 1, 3, 6, 12, and 24 hours were 30 (22), 31 (24), 32 (25), 34 (24), and 35 (28) Torr, respectively. MAPs did not correlate with birth weight, but they were lower among 19 stable infants < or = 26 w GA than among 17 stable infants > or = 27 w GA. History of preeclampsia, antenatal steroids, intratracheal epinephrine and cord hemoglobin did not influence MAP. Low 1-minute Apgar score and intracranial hemorrhage were associated with low MAP during the first day. CONCLUSIONS There is a wide variation of GA among ELBW infants. MAPs increase with GA and with postnatal age. Shortly after birth, arterial pressures are similar for stable and unstable infants. Failure to increase MAP between 3 and 6 hours of life should create concern. MAP < or = 28 Torr at 3 hours of life is a reasonable, but not absolute, predictor of the need for hypotension treatment.