Objective-To determine whether increased numbers of siblings and infec
tion in early life protect against allergic sensitisation. Design-Hist
orical cohort study. Setting-Sheffield, UK. Subjects-11 765 children a
ged 11-16 years for whom a history of neonatal infectious illness had
been recorded systematically at 1 month of age. Methods-A history of h
ay fever and family structure was obtained by postal questionnaire; ne
onatal illness history was ascertained from health visitor records; 72
3 children underwent skin prick testing with mixed grass pollen extrac
t.Results-The prevalence of hay fever was reduced (p<0.0001) among chi
ldren of younger mothers, and those from larger families. The number o
f older siblings exerted a stronger independent effect than the number
of younger siblings (p<0.001). Infants breast fed exclusively during
the first month were at higher risk (p<0.05) of subsequent hay fever,
independent of demographic factors. Adolescents at high risk of hay fe
ver by virtue of their family structure were more likely to be sensiti
sed to grass pollen (p<0.002). No significant relations emerged betwee
n hay fever and infection in the first month of life, even among child
ren born in June. Conclusions-The association of hay fever with family
structure is not due to reporting bias and reflects an environmental
influence on allergic sensitisation. The effects of sibship size, birt
h order, and infant feeding are consistent with a protective influence
of postnatal infection. The first month of life and the first postnat
al exposure to allergen are not the critical periods during which this
protective effect is determined.