Three studies were conducted to evaluate long-term memory longitudinally. In Study 1, 10-month-olds (N = 20) were taught to operate a toy in their homes and were tested at home after four months, as were age-matched (14 months) inexperienced controls (N = 20). Experienced infants were more willing to remain in the play situation, relearned faster than controls, and one operated the toy spontaneously. In Study 2, conducted 18 months thereafter, two subgroups (N = 5) of Study 1 groups and an age-matched (32 months) control group (N = 5) were observed in a lab playroom. Only the children with experiences at both 10 months and 14 months operated the toys without being shown. Children with a single 14-month experience made equivalent numbers of toy contacts and successful responses, however, and both groups exceeded controls. In Study 3, conducted 2+ years after Study 2, 36 children played in a novel playroom. Subgroups differed in amount and timing of experience (in Studies 1 and 2); a naive age-matched (60 months) control group (N = 6) was added. Controls took longer to make the toy work than children in the combined experience groups. Only experienced children elected to operate the toys later in the session. Two children verbally recalled part of the 10-month event. The findings are discussed in the light of their relevance to the assessment and description of memory during early childhood.