Recent theoretical and empirical study has considerably refined knowledge as to the dynamics of peer processes operative within secondary schools, but has tended to neglect models incorporating reciprocal effects. This situation results partly from a scarcity of appropriate data, and partly from limitations of older methodologies of estimating such models. This project makes use of rich survey data, including sociometric information and actual peer responses, and applies some recently developed estimation techniques to estimate the parameters, and the relative predictive power, of two alternative models for the influences of peers on academic achievement. The estimations are complicated by the appearance of unusual patterns in the data. However, they tend to indicate a better fit of the conceptualization which postulates direct reciprocal influences of each student's achievement on that of his peers, in contrast to an alternative model in which achievements depend only on the values of self and peers. In addition to the substantive results, the report also describes some methodological issues and comments on them. (Author/CTM)