The Bengal Basin is located in the foreland of the eastern Himalaya and records late Palaeocene to Pleistocene sedimentation. This makes the Bengal Basin an ideal place to study the evolution of the eastern Himalaya by provenance researches. Even though there have been several studies about its provenance, a consensus has not been reached about when the Himalayan and Gangdese‐derived detritus started to deposit in the eastern Himalayan foreland. Here, we study the provenance of Eocene to Pleistocene sedimentary successions of the Bengal Basin by detrital zircon U–Pb geochronology and Hf isotope analyses. Detrital zircon U–Pb age spectra show a predominant peak of ~470–580 Ma in all the samples. A peak at this time is prominent in sediments of the Tethyan Himalaya and Lesser Himalaya, which means the detritus eroded from Himalaya may have been deposited in the Bengal Basin after the late Palaeocene‐early Eocene, synchronous with detritus derived from the western and central Himalaya. Gangdese‐derived zircons were firstly recorded in the late Eocene Kopili Formation, and they were sporadic from the late Eocene to middle Miocene, which was probably due to the long distance between the Bengal Basin and the Lhasa Block, to the forebulge in the Lesser Himalayan Sequence, or to a wide tectonic separation at that time. The first influx of abundant Gangdese‐derived zircons occurred in the lower part of the Boka Bil Formation in the late Miocene (~10 Ma), which was probably caused either by the Brahmaputra River capturing the Yarlung Tsangpo River or by a much later terminal collision between India and Eurasia, or both.