This study sought to expand the modest literature investigating gene x envi
ronment interactions in the prediction of substance use. Our sample consist
ed of 591 male twins from the Minnesota Twin Family Study. Their relative g
enetic risk was estimated from their parents' substance-related diagnoses a
nd their environmental risk from their affiliations at age 11 with social g
roups likely to either encourage or discourage substance use. At age 14, th
e boys' own substance use was assessed. We hypothesized both main effects a
nd an interaction between our genetic- and environmental-risk variables in
the prediction of substance use by this young age. We further theorized tha
t the boys' inherited risk might take the form of temperament, specifically
externalizing tendencies. Using regression analyses and biometrical modeli
ng, we corroborated earlier research by finding evidence for a significant
interactive effect in the etiology of substance use. Our results suggest th
at low levels of environmental risk may buffer against the potentially unfa
vorable effects of high familial risk; however, when environmental risk is
high, the degree of familial risk is consequential. We were not able to sup
port our second hypothesis; rather, temperament predicted substance use onl
y through shared environmental factors.