Based on the empirical orthogonal function (EOF) analysis, the East Asia–Pacific (EAP) teleconnection is extrac-ted as the leading mode of the subseasonal variability over East Asia in summer, with a meridional tripole structure and significant periods of 10–30 and 50–70 days. A two-dimensional phase–space diagram is established for the EAP index and its time tendency so as to monitor the real-time state of EAP events. Based on the phase composite analy-sis, the general circulation anomalies first occur over the high-latitude area of Europe centered near Novaya Zemlya at the beginning of EAP events. These general circulation anomalies then influence rainfall over Northeast China, North China, and the region south of the Yangtze River valley (YRV) as the phases of EAP event progress. The rep-resentation, predictability, and prediction skill of the EAP teleconnection are examined in the two fully coupled sub-seasonal prediction systems of the Beijing Climate Center (BCC) and UK Met Office (UKMO GloSea5). Both mod-els are able to simulate the EAP meridional tripole over East Asia as the leading mode and its characteristics of evol-ution as well, except for the weaker precursors over Novaya Zemlya and an inconspicuous influence on precipitation over Northeast China. The actual prediction skill of the EAP teleconnection during May–September (MJJAS) is about 10 days in the BCC model and 15 days in the UKMO model based on correlation measures, but is higher when initialized from the EAP peak phases or when targeted on strong EAP scenarios. However, both of the ensemble pre-diction systems are under-dispersive and the predictable signals extend to 18 and 30 days in BCC and UKMO mod-els based on signal-to-error metrics, indicating that there may be further scope for enhancing the capability of these models for the EAP teleconnection prediction and the associated impacts studies.