Prediction skill regarding the Indo-Pacific climate has been rapidly enhanced in the past decade. While early prediction efforts were made based on statistical methods and/or simple climate models, recent climate predictions have been performed using comprehensive ocean–atmosphere general circulation models (OAGCMs). Both model performance of climate simulation and the data assimilation scheme have been improved to produce better prediction skill. Multimodel prediction results have been collected to gain higher skill, which is usually superior to that of the individual model. Most of the OAGCMs can now skillfully predict the Indian Ocean Dipole (IOD) at lead times of up to 1–2 seasons, and ENSO up to 6–9 months. Distinct SST patterns associated with different El Ni˜no flavors can also be well predicted at short-to-medium lead times. Furthermore, global climate anomalies induced by ENSO and IOD are realistically predicted. The subtropical dipole modes in the South Atlantic Ocean and the Indian Ocean (IO), the southern African climate, Asian monsoon precipitation, and Northern Hemisphere atmospheric circulation anomalies are predictable at short-to-medium lead times. Encouragingly, the JAMSTEC SINTEX-F model produces a useful skill of ENSO prediction at lead times of up to two years. And some strong IOD events can be well predicted up to one year ahead even if El Ni˜no's influence is suppressed. The results also suggest the importance of the IO–Pacific interbasin coupling and the recent global warming trend to climate predictability. For MJO prediction, the multimodel ensemble based on 12 OAGCMs achieves a useful skill (>0.5) of up to 26–28 days in advance.