This study used multinomial logistic regression to examine relationships be
tween child-care experience (in the context of overall family functioning)
and preschool attachment. Attachment behavior was assessed at 36 months wit
h the Strange Situation, and A, B, C, and D attachment classifications were
assigned using the MacArthur coding system. Maternal sensitivity was the s
trongest predictor of preschool attachment classification. With respect to
child-care effects, as at 15 months, no child-care factors (quantity, quali
ty, or type) predicted, in and of themselves, attachment security at 36 mon
ths. However, I of 3 interactions involving child care that were detected a
t 15 months reemerged at 36 months: When maternal sensitivity was low, more
hours per week in care somewhat increased the risk of the insecure-ambival
ent (C) classification. There was significant but modest stability of attac
hment classifications from 15 to 36 months, especially for children with A
and C classifications.