Discriminant analysis was used to explore the influence of the relative ages of siblings on their dyadic interactions, and to explore which interaction behaviors might discriminate among families and among interaction situations. Six dyadic interaction situations of 30 minutes duration were observed among members of 12 normal families. The observations took place in the families' homes once a week for five consecutive weeks. The interaction situations consisted of (1) child-directed interaction with mother, (2) child-directed interaction with father, (3) child-directed interaction with older sibling, (4) mother-directed interaction, (5) father-directed interaction, or (6) sibling-directed interaction. The Dyadic Parent-Child Interaction Coding System was modified in order to record the interactions of the target child with mother, father, or older sibling. This modified coding system provided a frequency count of 34 discrete positive and negative behaviors which may occur between parent/sibling and target child during play. Observations were made by two researchers who independently and unobtrusively recorded behaviors at 1-minute intervals while subjects played with a standard set of toys. Results of the discriminant analysis indicated that interaction situations could be correctly classified 89 per cent of the time based on the linear combination of six behaviors: three parent/sibling behaviors (identified as Acknowledge, Unlabeled Praise, Descriptive/Reflective Question) and three child behaviors (identified as Compliance to Direct Command, Whine, and Child Change Activity). A combination of five behaviors correctly classified family groups 87 per cent of the time (Child Change Activity, Child Whine, Descriptive/Reflective Question, Unlabeled Praise and Child Laugh). Implications for future normative studies are discussed. (Author/RH)