In this study the stability over the first 13 months of life of measures of infant cardiac activity (heart period and heart-period variability), their relations with each other, and their relations with a continuous-variable index of infant-mother attachment were investigated. The indexes of cardiac activity changed in an orderly way with development (increasing heart-rate variability, decreasing heart rate). There were moderate to high intercorrelations among the cardiac measures, particularly those indexing heart-rate variability (i.e., vagal tone, heart-period variance, and heart-period range). Regression analyses showed that the measures of heart-rate variability at 3,6, and 9 months were significant predictors of the continuous-variable index of security. The higher the infants' heart-rate variability, the higher were their attachment insecurity scores. Analyses of whether the conventional secure/insecure classification was related to the early infant cardiac measures indicated that measures of heart-rate variability were significantly higher in the insecure children.Recently, researchers have investigated the relation between attachment classification and measures of temperament (Belsky & Rovine, 1987; Bretherton, CConnell, &Tracey, 1980;Miyake, Chen, & Campos, 1985). This research has demonstrated a relation between concurrent measures of temperament and attachment and suggests that temperament, assessed during early infancy, may be a reliable predictor of attachment classification. For example, Miyake et al. (1985) reported that infant irritability measured at 3 months of age was predictive of patterns of resistant behavior during reunion in the strange situation at 1 year. These findings suggest that temperament may contribute significantly to the prediction of the quality of attachment.Neurophysiological processes may underlie the construct of temperament and mediate aspects of the relation between temperament and attachment. Contemporary theoretical perspectives have conceptualized temperament as a psychobiological construct (e.g., Goldsmith & Campos, 1982;Rothbart & Derryberry, 1981;Thomas & Chess, 1977). For example, Rothbart and Derryberry defined temperament in terms of constitutional differences in physiological reactivity and self-regulation of reactivity. Thus, children's approach and avoidant behaviors, observed in the strange-situation procedure (Ainsworth, Ble-