The Asian monsoon has long been argued to be a product of the Himalaya–Tibetan Plateau, and simulation experiments have confirmed the key role of the Himalaya–Tibetan Plateau in transforming regional atmospheric and oceanic circulations. However, temporal constraints on the strengthening of the Asian monsoon inferred from foraminifer isotopic and faunal data and terrestrial climatic and ecological records are inconsistent with each other, which has obscured the tectonic-climatic linkage. In particular, discriminating the post-middle Miocene global cooling from the monsoon upwelling cooling is critical, but poorly understood due to the lack of adequate constraints for monsoonal upwelling. Here we present new middle to late Miocene biomarker-based reconstructions of sea-surface temperature (SST) for the western Arabian Sea. Our new SSTs capture a long-term ocean cooling since ca. 14.8 Ma and a major drop in SST in the period 11–10 Ma after which the SSTs reached similar values as the Holocene. The new SST record is consistent with planktonic foraminifer, siliceous biota, and geochemical tracer studies, suggestive of ocean cooling and high productivity associated with monsoonal upwelling. The 11–10 Ma ocean cooling is not clearly expressed in other tropical oceans, indicating that the ocean cooling in the western Arabian Sea is not a simple reflection of global cooling. We interpret the 11–10 Ma ocean cooling as representing the establishment of monsoonal upwelling in the western Arabian Sea, triggered by strong cyclonic activities as a result of the Neogene outward expansion of the Himalaya–Tibetan Plateau.