The δ18O of speleothem fluid inclusions can provide information about local climate and the atmospheric circulation if the fate of oxygen isotopes from precipitation to the stalagmite is well understood. At Villars cave, SW France, the present-day relationship between the δ18O of precipitation and the δ18O of cave drip water has been determined by previous monitoring studies. In this article, we examine the link between drip water and fluid inclusions in speleothems from the same cave, and investigate whether the fossil water in the stalagmites provides a record of the isotopic composition of precipitation.Replicated fluid inclusion isotope measurements (δD and δ18O) from two modern calcite samples, which had precipitated under the monitored stalactites, are in agreement with the corresponding contemporaneous drip waters. The drip water, in turn, represents average pluri-annual precipitation. This gives strong evidence that the isotopic composition of the infiltrating water is preserved in the pores of speleothem calcite, thus providing a record of average precipitation δ18O at the site.Furthermore, we measured isotope ratios in fluid inclusions in the stalagmite vil-stm1, a 2300 year old sample with a high growth rate (up to 0.6 mm yr−1) and a high water content, which allows a high-resolution reconstruction of the drip water isotopic composition covering the last two millennia. The stalagmite is well dated by 13 U–Th ages and the counting of annual growth layers. The fluid inclusion δ18O varies by 2‰, with high δ18O values in Medieval times, and low values during the Little Ice Age, indicating a temperature influence on precipitation δ18O. However, temperature change alone cannot account for the full range of variability observed.Other proxies measured in the same stalagmite indicate that the cave environment has undergone substantial changes during this period: Stable isotopes and trace element concentrations in calcite show large variations, with amplitudes similar to those found in other Villars speleothems during large climatic shifts such as Dansgaard–Oeschger events or glacial–interglacial transitions. In the context of the last 2000 years, the proxy variability cannot be attributed to climate alone, but is likely influenced by anthropogenic deforestation and the associated changes in carbon dynamics and hydrology.These changes might also have an influence on the isotope ratios in drip water relative to precipitation, e.g. through an altered evaporation–transpiration ratio. Nevertheless, despite some common trends in all proxy records, the marked shifts in the other proxies are not seen in the fluid inclusion stable isotopes, which gives evidence that the relationship between precipitation and drip water δ18O in the past was similar to present-day conditions, and not much affected by the anthropogenic impact. A replication of the fluid inclusion isotope record using stalagmites from the same region will be necessary to determine whether local cave processes or large-scale temperature and atmospheric circulation dominate the fluid inclusion δ18O signal, and in how far the fluid inclusions can provide a direct record of precipitation δ18O.