This study collates linguistic, genetic, and archaeological data relevant to the problem of the IE homeland and proto-IE (PIE) migrations. The idea of a proto-Anatolian (PA) migration from the steppe to Anatolia via the Balkans is refuted by linguistic, archaeological, and genetic facts, whereas the alternative scenario, postulating the Indo-Uralic homeland in the area east of the Caspian Sea, is the most plausible. The divergence between proto-Uralians and PIEs is mirrored by the cultural dichotomy between Kelteminar and the early farming societies in southern Turkmenia and northern Iran. From their first homeland the early PIEs moved to their second homeland in the Near East, where early PIE split into PA and late PIE. Three migration routes from the Near East to the steppe across the Caucasus can be tentatively reconstructed - two early (Khvalynsk and Darkveti-Meshoko), and one later (Maykop). The early eastern route (Khvalynsk), supported mostly by genetic data, may have been taken by Indo-Hittites. The western and the central routes (Darkveti-Meshoko and Maykop), while agreeing with archaeological and linguistic evidence, suggest that late PIE could have been adopted by the steppe people without biological admixture. After that, the steppe became the third and last PIE homeland, from whence all filial IE dialects except Anatolian spread in various directions, one of them being to the Balkans and eventually to Anatolia and the southern Caucasus, thus closing the circle of counterclockwise IE migrations around the Black Sea.