With one exception, all previous studies of reconciliation in non-human primates (friendly reunion between former opponents) have focused on demonstrating the immediate, short-term effects despite the widely held view that reconciliation has a long-term function of repairing social relationships following aggression. To investigate this long-term function I compared mean interaction rates between opponents during the 10 d following reconciled and non-reconciled conflicts to baseline levels of interaction. Aggression rates during the 10 d after non-reconciled conflicts were significantly higher than the baseline rate, whereas after reconciled conflicts aggression was minimal. Similarly, grooming, proximity and approach rates during the 10 d after non-reconciled conflicts were significantly lower than the baseline rate whereas grooming, proximity and approach rates in the 10 d after reconciled conflicts were restored to baseline levels. These results indicate that there are consequences to not reconciling with a former opponent and highlight the fact that these may be costly in terms of increased risk of long-term aggression and reduced affiliation. The data support predictions from the Relationship-Repair Hypothesis suggesting that reconciliation functions as a mechanism for the repair of social relationships damaged by aggression.