Drug prevention in schools is a top priority in most Western countries and several well-designed studies have shown that prevention programs have the potential of reducing drug use in adolescents. However, most prevention programs are not effective and there are no general criteria available for deciding which program is effective and which is not. In this systematic review of the literature, the current scientific knowledge about which characteristics determine the effectiveness of drug prevention programs is examined. Three types of studies are reviewed: meta-analyses (3 studies were included), studies examining mediating variables of interventions (6 studies), and studies directly comparing prevention programs with or without specific characteristics (4 studies on boosters, 12 on peer-versus adult-led programs, and 5 on adding community interventions to school programs). Seven evidence-based quality criteria were formulated: the effects of a program should have been proven; interactive delivery methods are superior; the "social influence model" is the best we have; focus on norms, commitment not to use, and intentions not to use; adding community interventions increases effects; the use of peer leaders is better; and adding life skills to programs may strengthen effects.