This study examined whether family structure was associated with adolescent risk behaviors, including smoking and drinking. Family living arrangements have become increasingly diverse, yet research on adolescent risk behaviors has typically relied on measures of family structure that do not adequately capture this diversity. Data from the 1994-95 National Longitudinal Study of Adolescent Health were used to conduct logistic regression analyses that revealed adolescents in two biological married parent families were least likely to smoke or drink, whereas adolescents in cohabiting stepfamilies were most likely. Those in single-mother families and married stepfamilies were in between. Maternal socialization was related to reduced odds of smoking and drinking. Maternal modeling was positively associated with smoking and drinking. Family structure is indicative of distinct family processes that are linked to risky behaviors among adolescents.