Indian Ocean decadal sea-level variability is an active research area, with many unresolved questions due to inadequate observational coverage. In this study, we analyse 26 Coupled Model Intercomparison Project (CMIP) pre-industrial simulations and isolate two consistent modes of Indian Ocean variability, which collectively explain about 50% of the total decadal sea-level variance. With opposite sea-level signals in the southwestern and eastern Indian Ocean, the first mode is related to decadal modulation of the Indian Ocean Dipole (DIOD) and equatorial wind-driven dynamics. Though IOD is more independent of the El Niño–Southern Oscillation (ENSO) at decadal (r ~ 0.4) than interannual (r ~ 0.6) timescales, the DIOD–ENSO co-variability yields sea-level signals along the west coast of Australia, transmitted from the western Pacific via the Indonesian Throughflow. The second mode encompasses variability in the south Indian Ocean (SIODV), exhibiting a broad monopolar sea-level pattern east of Madagascar. In about half of the models, the SIODV is largely independent from DIOD (and decadal ENSO) and driven by south Indian Ocean wind-stress curl associated with meridional shifts in the Mascarene High (MH). In the other models, the SIODV lags the DIOD about 3 years. In those models, in addition to MH forcing, the DIOD-related alongshore wind stress off the northwest Australian coast triggers Rossby waves that also contribute to the SIODV, further west. The DIOD and MH forcing are mutually independent (r ~ 0.2). The results are broadly consistent with sea-level variations derived from the short altimeter data, despite an underestimation of the Oceanic bridge signals in CMIP models.