This study further investigates the 18-month and 3-year variability modes in the Indian Ocean and their possible use for the prediction of the Indian Ocean dipole (IOD). We argue that the IOD is not an isolated event triggered by a particular ocean–atmosphere state but a continuous oscillatory process, so that the extremely shallow thermocline in the eastern Indian Ocean during IOD events can be largely described (given our current observational time series of 20 years) as the constructive interference of the 3-year and 18-month signals. We propose to use this interpretation as a prediction tool of IOD events.