A method and system for mitigating the impact of latency in the initiation of a packet-based real-time media session. When an initiating station is setting up a requested session, the initiating station will generate and play out a repetitive tone pattern (e.g., ring tone) for the initiating user to hear. The tone pattern is predefined in the initiating station, rather than being received from the network during session setup. The repetitive tone pattern preferably relaxes the initiating user and reduces or prevents "dead air" during session initiation.