Time-series O isotope profiles for three U–Th dated stalagmites have revealed that for much of the Holocene, a site on the Atlantic seaboard (SW Ireland) exhibits first-order δ18O trends that are almost exactly out of phase with coupled δ18O curves from two southern European sites (SE France and NW Italy). In the Irish stalagmite (CC3 from Crag Cave, SW Ireland), low δ18O at 10,000calyr BP reflects cool conditions. By the early to mid-Holocene (9000–6000calyr BP) δ18O had increased, reflecting the onset of warmer conditions on the Atlantic seaboard. This shift to higher δ18O was accompanied by a marked increase in the stalagmite extension rate, reinforcing our interpretation that this was a period of relative warmth. Except for an episode of increased extension rate about 5500yr ago, δ18O in the Crag stalagmite exhibits a gradual decrease, accompanied by declining extension rates between 7800 and 3500calyr BP, interpreted as a cooling trend. There is evidence for increases in both δ18O and stalagmite extension rate in the period from 3500calyr BP to the present day suggesting a return to warmer conditions on the Atlantic seaboard. In the stalagmite from NW Italy (ER76, Grotta di Ernesto, Trentino province) the early-Holocene (c. 9200-7800calyr BP) is characterised by high δ18O, probably indicative of warm and/or dry conditions. Exceptionally low δ18O from 7800 to 6900calyr BP at this site reflects a well-defined wet phase (Cerin wet phase). In the last three millennia, this stalagmite exhibits a shift to lower δ18O, interpreted as some combination of cooler and/or wetter conditions. Unlike the Irish stalagmite, the Italian sample does not show a correlation between δ18O and extension rate. Instead, its extension rate correlates roughly with δ13C, presumably reflecting a climate-driven vegetation change. In the early Holocene, δ18O in the French stalagmite (CL26, Grotte de Clamouse, Herault province, SE France) was low relative to its Holocene average. For much of the period since c. 3500calyr BP this stalagmite exhibits higher δ18O than in the early Holocene, suggesting warmer conditions. Like the Irish stalagmite, the French sample exhibits a well-defined correlation between δ18O and extension rate. Had drip-water availability been the dominant control on δ18O at this semi-arid site then higher δ18O would have been accompanied by lower, not higher extension rates. This suggests strongly that temperature rather than rainfall amount was the dominant control at this site. While conclusions regarding the patterns of climate variability on a continent scale must remain tentative because of the limited number of stalagmites studied we argue that early Holocene warm conditions on the Atlantic seaboard (Irish site) coincided with relatively cool conditions at the Clamouse site. By c. 3500yr ago the pattern appears to have been reversed.