Suicide and accidents are the leading causes of death among American Indian (AI) adolescents. Engaging in health-compromising behaviors (HCB) is higher among AI youth than among multicultural, national samples of adolescents. These HCBs include: smoking, drinking alcohol, drug use, and delinquency. Studies that identify legitimate predictors of these behaviors among AI adolescents are needed to guide research and interventions. Primary socialization theory (PST) suggests that peer groups, family, and school are the only areas where adolescents are directly taught to accept or reject deviant or normative behavior. Gateway theory indicates that use of certain drugs by adolescents, such as cigarettes or alcohol leads to the use of additional illicit drugs. Both of these theories were investigated in the current study as possible guides to identifying risk factors for HCBs among AI adolescents. The behaviors investigated in this study were alcohol use, cigarette use, illicit drug use, delinquency, suicidality (i.e., ideation and behaviors), and self-protection (seatbelt and helmet use) at Time 2. Predictor variables included behaviors and intrapersonal factors at Time 1 (one year earlier). All variables came from measurements provided by the National Longitudinal Study of Adolescent Health. Multiple linear regressions were calculated for all youth together, males only, and females only to determine which combination of predictors accounted for the most variance in the target behavior. Support was found for PST across behaviors in that variables measuring the primary socialization sources (i.e., peer groups, family members, and involvement with school) were significantly predictive ofHCBs one year later in all regressions calculated. Little support was found for gateway theory regarding substance use, as experimentation with alcohol and cigarettes at Time 1 was not predictive of illicit drug use at Time 2. There were 398 self-identified AI adolescents at Time 1, and 298 at Time 2, included in this study. There were 175 females and 123 males, ranging in age from 13 to 20 at Time 2. One limitation of this study is that all information was obtained via selfreport. Other limitations, implications for future research, and areas for prevention or intervention with AI youth are discussed.