Serial measurements of urinary arginine vasopressin (AVP) were made in six severely birth asphyxiated newborn infants. In five infants serial plasma concentrations were also evaluated. There was a strong negative correlation between plasma AVP and plasma osmolality in these infants (r = -0.52, P = 0.0012). In neither the individual babies nor the group as a whole was there a significant correlation between plasma AVP and the urinary excretion of AVP even if the latter was standardised for creatinine content. Normal development at follow up was only observed in two asphyxiated infants who had consistently low urinary arginine vasopressin levels in the first days of life. Infants with consistently high urinary vasopressin concentrations either died or were severely abnormal in their subsequent development.