Predictors of fathers' involvement in childrearing activities were investigated as part of a study of the ecology of urban childrearing in five western societies. Selected on the basis of a stratified random sample technique, participants were 96 white fathers from intact families having a 3-year-old child. Network theory provided three hypotheses regarding how a father's social network might affect his involvement in childrearing. A variable was constructed for each hypothesis; these variables were "percent male network," a measure of the percentage of the exchange/activity network who are male; "network range," a measure of the size and diversity of the social network; and "non-kin intensity," a measure of the degree of choice operative in social network selection and involvement. The index of childrearing involvement was labeled "childrearing advice." The review also provided a causal path hypothesis in which social network was proposed as a link mediating the effect of education on childrearing involvement. Data were gathered in social network interviews designed to generate a multifaceted picture of parents' social worlds; total network membership was obtained on the basis of results on a series of role context probes. To analyze the data, the three network variables were entered into a multiple regession along with a set of eight demographic and situational variables, including father's work hours, mother's work hours, sex of child, swing shift (dummy variable), father's education, occupational status, family ethnicity, and father's age. Only non-kin intensity strongly predicted childrearing advice, and the causal path hypothesis was confirmed. (RH)