The effects of premature birth on attachment have generally been examined from the infant's perspective. There is a lack of data concerning parental attachment representations toward a premature child. Because of the psychological stress engendered in parents confronted with a premature birth, we hypothesized that their attachment representations would be altered during the first months after the hospital discharge. Fifty families with a premature infant (25-33 gestation weeks) and a control group of 30 families with a full-term infant participated to the study. Perinatal risks were evaluated during hospitalization. To assess mothers' representations of their infant, the Working Model of the Child Interview (WMCI, Zeanah & Benoit, 1995 & Benoit, Zeanah, Parker, Nicholson, & Coolbear, 1997) were administered when their children were 6 and 18 months old. The severity of the perinatal risks was found to have an impact on the mothers' attachment representations. At six months, only 20% of the mothers of a prematurely born infant (30% at 18 months) had secure attachment representations, vs. 53% for the control group (57% at 18 months). Furthermore, mothers of low-risk premature infants more often had disengaged representations, whereas distorted representations were more frequent in the high-risk group of premature children. These findings suggest that the parental response to a premature birth is linked to the severity of postnatal risks. The fact that secure attachment representations are affected in mothers of low-risk infants just as much as they are in mothers of high-risk infants points to the need to conduct further studies aimed at evaluating whether preventive intervention for both low-risk and high-risk premature will be helpful.