The lake level of the Dead Sea, Southern Levant, has fluctuated with an amplitude of ∼250 m in response to the last glacial-interglacial cycle. This exceptional sensitivity to climate change, and the availability of long sedimentary archives, make the Dead Sea a benchmark for long quantitative paleohydrological reconstructions. However, discontinuities and chronological uncertainties in the marginal sedimentary record have hampered the reconstruction of Dead Sea lake levels beyond the Last Glacial (70–14 ka before present, BP). Here, we apply a two-pronged methodology. First, we measure the lake water density along ICDP deep core 5017-1-A using a new method, Brillouin spectroscopy on two-phase halite fluid inclusions; we combine it with the composition of pore water and the thickness of halite layers in the core to reconstruct lake level, volume, mass balance and subsidence rate. Second, we tune the chronology of lake levels from outcrops by matching it to the chronology of the deep core. The resulting lake level reconstruction, spanning 237–70 ka BP, is validated by the excellent agreement between outcrop- and mass balance-based methodologies. It shows a long-term recession of the lake, its level decreasing from one interglacial to the other, down to a Holocene record low. There are two reasons for this lake level fall. First, with an average rate of 2.65 ± 0.15 m/ka, subsidence has outpaced sedimentation at least over the last ∼130 ka. Second, by reducing the solute inventory of the lake, massive halite precipitation events such as that of 131–116 ka BP have durably increased surface water activity and evaporation, and thus lowered the lake level, up to today. Conversely, our analysis suggests that, during 191–11 ka BP, the dissolution of Mount Sedom salt diapir and freshwater inflows provided to the lake about three times the mass of solute NaCl contained in the modern Dead Sea (in 1985). This massive solute influx, occurring mainly during glacial highstands, strongly contributed to lowering surface water activity and evaporation and, therefore, to increasing the lake volume. Our results suggest that Dead Sea lake levels are more accurately interpreted in terms of climatic change if surface water activity is taken into account.