A recent trend in the study of forgiveness is to look at forgiveness outside of traditional victim-perpetrator dyads. One way of going beyond these dyads is to look at third-party forgiveness. A recent advance in the study of relationship valence is to look at the amount of positivity and negativity in a relationship as on independent scales rather than on two ends of the same spectrum. This allows for categorization of relationships that are high in both positivity and negativity—ambivalent relationships. This study attempts to combine these two recent advances. I hypothesized that participants would have more difficulty forgiving offenders with whom they had ambivalent relationships, that participants would be more forgiving when they were personally offended compared to when they were a third-party to an offense committed against a friend, and that participants would be more forgiving of a perpetrator when an offense was committed against an ambivalent relation than when the same offense was committed against a positive relation. Strong support was obtained for the first hypothesis, but little support for the second and third hypotheses. The implications of these results are discussed and possible directions for future research are recommended.