OBJECTIVE: To evaluate the efficacy of the clinical neurologic examination, hematocrit and glycemia in the diagnosis of periventricular-intraventricular hemorrhage in the newborn weighing less than 2000g considering the neurosonography as the gold standard.METHODS: This is a cohort prospective study from May 18th, 1994 to May 17th, 1995 carried out at the Hospital das Clinicas da UFMG. The study group comprised 38 newborns with the ultrasound diagnosis of periventricular-intraventricular hemorrhage; the control group comprised 81 newborns who although submitted to the same evaluation protocol did not show any echographic signs of hemorrhage. The ultrasound examinations were all done by the same researcher who was not aware of the clinical history or the neurologic and laboratory examinations.RESULTS: Clinical neurologic alterations, mainly generalized hypotonia, were found to be associated with hemorrhage (p = 0.009), but there was a confounding effect due to the gestational age of the newborn. The predictive values of a positive or negative test were only 45% and 79%, respectively, which is clearly unreliable for the diagnosis of the hemorrhage. Hematocrit and glycemia were not clinically relevant either.CONCLUSIONS: The clinical neurologic examination is not reliable to indicate periventricular-intraventricular hemorrhage in the newborn. Newborns at risk should be routinely screened by neurosonography.