One-hundred-and-seventy infants resident in the Oxford Record Linkage Area and dying suddenly and unexpectedly in the five-year period 1966-70 were linked with the records concerning their delivery. For each of these cases three controls were chosen, matched for maternal age, parity, social class, legitimacy, and place and year of delivery. The files of general hospital admissions were scanned to identify admissions to the mothers during the pregnancy, and admissions to the infants, for both index cases and controls.
Significant associations were found with short gestation, low birthweight, slight growth retardation, twin delivery, `insult' during pregnancy, induction of labour, neonatal jaundice, congenital defects, subsequent hospital admission, and month of birth. There was no association with breast feeding but some evidence for contact with other cases during hospital admission.