Part of a larger study of adolescent social-cognitive development and its influence on mother/child interaction, this investigation compared three methods of recording observations of mother/child interaction. It also explored correlations between sociodemographic variables and factor scores that summarized interaction variables. The sample consisted of 128 single adolescent mothers, ages 15 to 20, with a firstborn child from 9 to 27 months of age. Subjects, 93 percent of whom were black, averaged 10.5 years of school. Observations of mother/child interaction occurred in the home during a developmentally appropriate teaching task and a period of free play. Four observers were trained for 12 to 15 hours before the project and received a refresher course two-thirds of the way through the data collection. Interobserver reliability was measured at the midpoint and at the end of the data collection year according to two methods. Scores on two of the observation scales were standardized to make them comparable and to remove the bias of systematic scale preference. The methods for recording behavior (Barnard's Teaching Scale, Schaefer's Attachment Inventory, and Clarke-Stewart's Rating Scale), required different degrees of inference by observers and examined different-sized units of data. Results of factor analyses of the child and maternal data were reported separately for each of the instruments and then comparatively for all three instruments. Few sociodemographic variables were correlated with instrument factors or items. Conclusions discussed the overlapping relationships between the scales and the contribution of each to efforts to describe maternal/child interaction. (CB)