Past studies have shown that people favor their own group members when they themselves are the third-party punisher. But in real life, most people do not have adequate power to enforce justice by themselves. Instead, they are most likely to enforce moral norms by choosing who to grant the power to punish (through election, petition, public discussion, etc.), and such choice relies crucially on their evaluation of the third-party punishment and the punisher. Do children have group bias when evaluating other’s moral punishment at its emergence? In this study (N = 197, age 3-7), we found that in evaluating other’s third-party punishment, children showed in-group favoritism toward in-group perpetrators and out-group discrimination toward out-group perpetrators, evaluating a third-party punishment more positively when it is enforced on an out-group perpetrator than an in-group perpetrator. However, the group affiliation of the victim did not induce biased evaluation for a third-party punishment. In this Chinese sample, we also found that children are biased in judging the deserved punishment of a transgression and in their third-party punishment behavior. They favored in-group perpetrators, which was in accordance with previous study conducted on western samples. We also found that over development, children judged a transgression as more serious and deserve more punishment, and gave increasingly higher evaluation toward other’s third-party punishment. Taken together, this study added to the evidence that group bias is rooted deeply in ontology, and punishment toward injustice is biased even at its emergence.