This paper describes the specification and testing of a model of protective parent and peer factors in peer refusal skills. Two modifiable protective factors suggested by relevant research on adolescent substance use-child attachment with parents and association with peers having prosocial norms-were incorporated as independent variables in the model. The effects of parent and child attendance at skills training interventions were also assessed. Covariance structure modeling of data from a sample of 209 families participating in a controlled study of a family-oriented skills intervention was used to test two versions of the model, one version addressing attachment and skills training attendance specific to mothers and one specific to fathers. Following two indicated modifications of the original model, strong fits with the data were achieved for both mother and father versions of the model; hypothesized protective factor effects and skills training effects were significant.