Thirty-three families, each with a premature infant born less than 33 gestational weeks, were observed in a longitudinal exploratory study. Infants were recruited in a neonatal intensive care unit, and follow-up visits took place at 4 months and 12 months of corrected age. The severity of the perinatal problems was evaluated using the Perinatal Risk Inventory (PERI; A.P. Scheiner & M.E. Sexton, 1991). At 4 months, mother infant play interaction was observed and coded according to the CARE-index (P.M. Crittenden, 2003); at 12 months, the Strange Situation Procedure (SSP; M.D.S. Ainsworth, M.C. Blehar, E. Waters. & S. Wall, 1978) was administered. Results indicate a strong correlation between the severity of perinatal problems and the quality of attachment at 12 months. Based on the PERI, infants with high medical risks more frequently tended to be insecurely attached. There also was a significant correlation between insecure attachment and dyadic play interaction at 4 months (i.e., maternal controlling behavior and infant compulsive compliance). Moreover, specific dyadic interactive patterns could be identified as protective or as risk factors regarding the quality of attachment. Considering that attachment may have long-term influence on child development, these results underline the need for particular attention to risk factors regarding attachment among premature infants.