Sediment profiles from several lakes/ponds in the northern, ice-free area of James Ross Island, Antarctica, were the subject of a multi-disciplinary palaeoenvironmental investigation. The results, obtained from lithological, geomagnetic, geochemical and diatom analyses, and the frequency of Branchinecta eggs, were evaluated with multivariate statistics and provide a fairly detailed picture of climate change during the last 5000 years. New radiocarbon dates are combined with previously published 14C dates, so as to date the palaeoenvironmental and palaeohydrological changes identified by the stratigraphic studies. In combination with the findings of glacial geological studies our data suggest that the Brandy Bay glacier began to recede rapidly just before 5000 yr B.P., due to calving and moisture starvation caused by cold and arid conditions. In one of the ponds studied gypsum began to precipitate as a consequence of these conditions when the influence of glacial meltwater ceased. This hypothesis was tested by hydrochemical modeling. At approximately 4200 yr B.P. the climate became more humid (and warmer) and the pond supported a fairly diverse biota, which included a variety of diatom species and, for example, Branchinecta gaini, a crustacean that feeds on benthic cyanobacterial mats. The Brandy Bay glacier probably advanced during this humid period, but the advance was interrupted by subsequent arid conditions which started at 3000 yr B.P. During this arid phase, glaciers became completely absent from the study area. As a result formerly large lakes, fed by glacial meltwater, transformed into small, enclosed, brackish water bodies that contained B. gaini. The arid and fairly cold conditions lasted for 1500–2000 years, until approximately 1200 yr B.P. when the aquatic systems expanded and their salinity was lowered as a result of increased snow accumulation and glacial expansion. The increased deposition of minerogenic matter caused by glacial erosion and meltwater inflow probably caused the disappearance of Branchinecta on James Ross Island. We regard this most recent phase as a period of increased humidity and warmth compared to the previous period, although the conditions appear to be less warm and humid than those which prevailed during the climatic optimum between 4200 and 3000 yr B.P. In spite of certain correlation problems we tentatively correlate our results to other studies of Holocene environments in Antarctica, and from that it can be suggested that the climatic optimum was of circumpolar significance. Therefore the climatic oscillations recorded in sediments on James Ross Island may reflect oscillations in the anticyclonic stormtracks and the strength of the high pressure cell over the inland ice sheet.