Seventy 15-month-old infants were studied at home before starting child care, during adaptation (mothers present) and separation (first 9 days without mothers) phases, and 5 months later. Security of infant-mother attachment was assessed before and 3 months after child care began. In the separation phase, salivary cortisol rose over the first 60 min following the mothers' departures to levels that were 75% to 100% higher than at home. Compared with insecure infants, secure infants had markedly lower cortisol levels during the adaptation phase and higher fuss and cry levels during the separation phase, and their fuss and cry levels were significantly correlated with their cortisol levels. Attachments remained secure or became secure if mothers spent more days adapting their children to child care.