In this study, we investigate the predictability of the tropical Indian Ocean (TIO) sea surface temperature anomalies (SSTA) using the recently released North American Multimodel Ensemble dataset (NMME). We place emphasis on the predictability of two interannual variability modes: the Indian Ocean Basin mode (IOBM) and the Indian Ocean Dipole (IOD). If defined by a 0.5 correlation skill, we find that the statistically skilful predictions correspond to an approximately 9- and 4-month lead for the two modes, respectively. We then applied a newly-developed predictability framework, i.e. Average Predictability Time method (APT), to explore the most potentially predictable mode (APT1) for the TIO SSTA. The derived APT1s correlate significantly to the IOBM and IOD, but are also characterised by several significant differences, which implies that there is a close link between the variability-related modes and the predictability-defined modes. Further analysis reveals that the predictability source of the IOBM-related APT1 originates from ENSO-induced thermocline variation over the southwest Indian Ocean, whereas wind-driven upwelling near Sumatra dominates IOD-related APT1. This study provides insights into the understanding of TIO SSTA predictability and offers a practical approach to obtain predictable targets to improve the TIO seasonal prediction skill.