El Niño and Southern Oscillation (ENSO) events are usually monitored bytropical Pacific sea surface temperature anomaly (SSTA) patterns with dra-matic impacts on the global climate. To explore the diversity of the tropicalPacific SSTA, a novel method combined empirical orthogonal function analysi sand the K-means clustering algorithm is carried out to classify SSTA patternsduring 1950–2016. Meanwhile, the total distance variance and total silhouettevalue are introduced to determine an optimal number of distinguishable repre-sentative SSTA patterns. Ten SSTA patterns are obtained, which shows fre-quent basin-wide warming and extreme cold ENSO events in recent decades.It may be attributed to the changes in composition of the intrinsic modes alongwith the background mode of slowly increasing east–west SST gradient. Thecomparative analysis between periods 1950–1969 and 1997–2016 suggests thatthe two regimes of tropical Pacific SSTA, featured as extreme warm and mod-erate warm/extreme cold patterns respectively, become more distinct underrecent global warming.