The relation between intraventricular haemorrhage (IVH) and hyaline membrane disease (HMD) was studied in singletons that came to necropsy at Hammersmith Hospital over the years 1966-73. The incidence of IVH in singleton live births was 3-22/1000 and of HMD 4-44/1000. Although the high figures were partily due to the large number of low birthweight infants born at this hospital, the incidence of IVH in babies weighing 1001-1500 g was three times as great as that reported in the 1658 British Perinatal Mortality Survey. Most IVH deaths were in babies with HMD, but the higher frequency of IVH was not associated with any prolongation of survival time of babies who died with HMD as compared with the 1958 survey. IVH was seen frequently at gestations of up to 36 weeks in babies with HMD but was rare above 30 weeks' gestation in babies without HMD. This indicated that factors associated with HMD must cause most cases of IVH seen at gestations above 30 weeks. Comparison of clinical details in infants with HMD who died with or without IVH (at gestations of 30-37 weeks) showed no significant differences between the groups other than a high incidence of fits and greater use of alkali therapy in the babies with IVH. During the 12 hours when most alkali therapy was given, babies dying with IVD received a mean total alkali dosage of 10-21 mmol/kg and those dying without IVH 6-34 mmol/kg (P less than 0-001). There was no difference in severity of hypoxia or of metabolic acidosis between the 2 groups. Babies who died with HMD and germinal layer haemorrhage (GLH) without IVH had received significantly more alkali than those who died with HMD alone, whereas survivors of severe respiratory distress syndrome had received lower alkali doses than other groups. It is suggested that the greatly increased death rate from IVH in babies with HMD indicates some alteration of management of HMD (since 1958) as a causative factor. Liberal use of hypertonic alkali solutions is the common factor which distinguishes babies dying with GLH and IVH from other groups of babies with HMD. Although the causal nature of this association remains unproved, it seems justifiable to lrge caution in alkali usage.