In a sample of established working- and middle-class families with school-aged children (N= 307 wives and 307 husbands), neither husbands’ nor wives’ testosterone showed a direct connection with marital quality. In contrast, the association between husbands’ testosterone and positive and negative marital quality (as evaluated by both spouses) was conditional on husbands’ role overload. When perceptions of role overload were elevated, higher testosterone levels were associated with lower levels of marital quality. When perceptions of role overload were low, higher testosterone was linked to greater marital quality. The study supports the biosocial model such that, depending on perceptions of the social context, testosterone enables positive behavior in some instances and negative behavior in others.