Although it is common for secondary schools to implement alcohol policies to reduce alcohol misuse, there has been little evaluation of the efficacy of these policies. The purpose of this study was to test the impact of the degree and type of alcohol policy enforcement in state representative samples of secondary students in Washington State, USA, and Victoria, Australia (n = 1848). Multivariate logistic regressions were used to examine the prospective association between student reports of school alcohol policy in Grade 8 and self-reported alcohol use in Grade 9, controlling for age, gender, state, family socio-economic status and Grade 8 alcohol use. The likelihood of students drinking on school grounds was increased when students perceived lax policy enforcement. Student perceptions of harm minimization alcohol messages, abstinence alcohol messages and counselling for alcohol policy violators predicted reduced likelihood of binge drinking. Students perceiving harm minimization messages and counselling for alcohol policy violators had a reduced likelihood of experiencing alcohol-related harms. Perceptions of harsh penalties were unrelated to drinking behaviour. These results suggest that perceived policy enforcement may lessen drinking at school 1 year later and that harm minimization messages and counselling approaches may also lessen harmful drinking behaviours as harm minimization advocates suggest.