Our study focused on an educational intervention designed to increase the self-esteem of low-achieving eighth graders. The intervention was a substance-specific life skills program built upon teacher-centered versus student-centered teaching methods. A cluster analysis identified four consumption groups prior to the intervention: A potentially curious cluster (PC), a currently curious cluster (CC), an experimenter cluster (EP) and a consumer cluster (CO). Self-esteem was measured three times in a pre-, post- and retention test design. The in terms of percentage biggest PC cluster only responded positively. A consistent increase was detected among both different teaching methods and both genders. Thus, self-esteem as an important psychological construct of physical well-being could be influenced positively by many students in the context of a substance-specific life skills program. Educational consequences to increase the efficiency of scholastic measures are discussed.