The article examines how direct and indirect intergroup friendships can moderate prejudice based on whether prejudice has predominantly affective or cognitive bases. Three studies found that: 1) Direct friendships more strongly reduced prejudice towards groups eliciting affective responses, while indirect friendships more strongly reduced prejudice towards cognitively-based groups. 2) Direct friendships reduced prejudice more for individuals with affective prejudice bases; indirect friendships reduced prejudice more for those with cognitive bases. 3) A third study of Catholics and Protestants replicated these moderation effects on measures of trust and action tendencies, providing further support. Intimate contact through both direct and indirect friendships can benefit intergroup relations.