<p>Lakes are sensitive to climate change and their sedimentary components play a pivotal role as environmental recorders. In the past, lacustrine carbonates have been utilized in a number of studies attempting at a quantitative reconstruction of rainfall and/or precipitation-evaporation changes based on the biogenic or bulk carbonate &#948;<sup>18</sup>O signature. While these studies are built on sound theoretical grounds of mass balance and kinetic isotopic fractionation, the challenge often overlooked is the mineralogically mixed nature of carbonates comprising the bulk.</p><p>We report a case study from Lake Van, the world&#8217;s largest alkaline lake. Our time series comprising ca. 140 ka documents not only changing proportions of surface water calcite and aragonite, but also diagenetic bottom-water dolomite and, for the first time in Lake Van, early diagenetic bottom-water aragonite. Importantly, in the Lake Van profile primary and early diagenetic carbonates (in particular aragonite) are concurrent rather than mutually exclusive. A comprehensive comparison of the &#948;<sup>18</sup>O and &#948;<sup>13</sup>C compositions of singled out water column, biogenic (ostracod valves) and diagenetic carbonates shows, that each of the fractions forms a distinctive cluster characteristic for the depth and timing of their formation. Also, the differences between &#948;<sup>18</sup>O values of concurrent deep-water carbonate phases exceed what is expected from mineral-specific fractionation. Our data suggest that, an uncritical and unchecked application of the isotopic composition of the bulk carbonate fraction in quantitative climate reconstructions can severely compromise the results. We also advocate that, among different carbonate fractions in Lake Van, monospecific biogenic samples most faithfully reflect the oxygen isotopic composition of the lake water contemporaneous to their deposition, while the carbon composition of biogenic samples is additionally influenced by the organism microhabitat.</p>