The quality of the attachment relationship between mother and infant is typically determined in the Strange Situation. The assignments of infants to the A (avoidant), B (secure), and C (resistant) attachment classes are largely but not exclusively based on measurements during the reunion episodes. In this paper, the measurements in the reunion episodes are used to derive a clustering of the infants via three-way mixture method of clustering, a technique especially designed for clustering three-way mixture method of clustering, a technique especially designed for clustering three-way data (here: infants, variables and episodes). The results are compared with the A-B-C classification, and the relevance of the outcomes for attachment research are discussed. At the same time, the paper aims to demonstrate the use and usefulness of the three-way clustering procedure for data from the social and behavioural sciences.