Disclosed methods of achieving load balancing among a plurality of border elements in a next generation networks (NGN) communication network include a method for processing and user endpoint originated messages (UEOM), such as a register or invite message, generated by a user endpoint. A work load information data structure indicative of a first border element's work load and a second border element's work load is maintained when the border elements exchange work load information. Based on the work load information, a first border element responds to a UEOM by either (1) accepting the UEOM for first border element processing, (2) determining whether to accept the UEOM for first border element processing or redirect the UEOM to a second border element for second border element processing, or (3) rejecting the UEOM.