This report concerns the similarities and differences between maternal and observer Attachment Q-Set (AQS) security scores for 41 preterm and 38 fullterm infants. The following variables were assessed: maternal sensitivity (8 months); parental stress (8 months); mother- and observer-derived AQS measures of attachment security and dependency (12 months); and infant temperament (18 months). "Strange situations" were recorded at 18 months. Previous analyses of these data had shown that although mother and observer AQS security scores were correlated (r = .55), only observer scores converged with the strange situation. Two additional series of analyses were carried out. The first revealed that AQS security scores of sensitive mothers, but not those of less sensitive mothers, are correlated with those of observers. Maternal sensitivity was unrelated to the lack of correspondence between mothers’ AQS scores and strange situation classification. Second, a factor analysis of the above variables revealed that while observers’ sensitivity and attachment security scores and mothers’ security scores loaded on to a Relationship Security factor, mothers’ security ratings also loaded on to an Infant Fussiness factor. This finding suggests that although both mothers and observers focus on items indicative of security when completing the AQS, mothers may place an additional emphasis on fussiness-related items. Further support for this suggestion was found in an analysis of covariance. When variance attributed to fussiness-related items of the AQS was partialed out of the mother-derived security score, the residual mapped on to the secure/nonsecure distinction in the strange situation. However, fussiness was unrelated to the mother-observer AQS correlation, indicating that different sources of variance may be involved in the mother-observer AQS correlation and in the correspondence between mothers’ AQS scores and the strange situation.