Traditional seasonal forecasting systems that permit skill assessment out to 12-month lead times show relatively high skill for predictions of sea level variations on the US West Coast (WC). The multi-season WC skill has been attributed to the influence of predictable ENSO signals and associated coastal wave dynamics. We examine US coastal sea level prediction skill in two extended seasonal prediction systems (24-month lead times) based on the CESM2 and E3SMv2.1 models. Potentially useful WC sea level skill extends into Year-2 of the hindcasts and exceeds that for the Niño3.4 sea surface temperature at long leadtimes. We discuss the mechanisms underpinning extended seasonal predictability of coastal sea level and their representation in these two prediction systems.