Parental attachment of 121 high-risk women, 61 partners of high-risk women, 182 low-risk women, and 117 partners of low-risk women, was studied at the first week postpartum and 8 months following birth. The tests of theoretical models showed low predictive ability explaining from zero to 21% of the variance in parental attachment in the four groups over the two test periods. Empirical respecified models predicting parent-infant attachment at the first week postpartum and 8 months explained 31% and 29% of the variance among high-risk women, 69% and 45% among high-risk partners, 41% and 53% among low-risk women, and 35% and 38% among low-risk partners. Parental competence was a major predictor of parental attachment over all test periods for all four groups. Early parent-infant contact following birth was never a predictor except at 8 months when, among low-risk women, the opposite effect than that expected was observed; the later women held their infants the higher was their attachment. High-risk women scored significantly higher than low-risk women during the first week postpartum only.