The present disclosure relates to the field of communications, and discloses a method for determining path pre-congestion. The method is applicable to an Ethernet path. Pre-congestion is determined and flagged through an ingress node or an intermediate node in the Ethernet path, and a pre-congestion status indication packet is sent by an egress node to the ingress node or a policy server, so that the ingress node or the policy server obtains pre-congestion information of the Ethernet path, and notify an application function entity AF of performing admission control. When the method in the present disclosure is used, problems of packet loss and excessively long time delay caused by the conventional congestion control technology may be avoided, thereby ensuring user experience of existing services.