AIMS: The aim of the study was to assess sociodemographic, parental and peer predictors of self-reported drunkenness in mid-adolescence. METHODS: The data were obtained from a national school survey covering a random half of all Icelandic 9th and 10th grade students (mean age 14.7 years). The overall response rate was 92% (n = 3,913). RESULTS: 30% of the respondents reported having been drunk or intoxicated at least once during the 30-day period prior to the survey. Multivariate logistic regression analysis showed that living with a single father, or in an urban area; having friends who get drunk or accept drunkenness; being supported by friends; having a father who gets drunk; having parents who assent to drunkenness; and lacking support from parents, were significantly associated with drunkenness. The strongest predictors of drunkenness were drunkenness among friends and friends' acceptance of drinking. The study did not find significant differences in the odds of drunkenness by gender or parental education. Students in 10th grade had higher odds of drunkeness than 9th grade students, which was accounted for by different family and peer contexts of younger and older adolescents. CONCLUSIONS: Residence, family structure, high peer support, peer acceptance, peer drunkenness, parental acceptance, father drunkenness, and low parental support was related to higher odds of drunkenness in mid-adolescents. The results give directions to future research and interventions intended to prevent alcohol abuse in this population.