Loess in the Carpathian Basin is some of the thickest and most complete in Europe. Located in the Vojvodina region of the southern Carpathian Basin the Crvenka loess-palaeosol section appears to preserve a detailed climate proxy archive of the last glacial–interglacial cycle. Central to the interpretation of the site is a detailed and independent age model. Here, the results of detailed optically stimulated luminescence (OSL) dating and elevated temperature post-IR infrared stimulated luminescence (post-IR IRSL) dating are presented. Quartz OSL ages appear accurate to about 50–60 ka, where 2D0 values are reached, while elevated temperature post-IR IRSL yields more accurate ages below this. In line with recent results, the latter signal appears to show negligible fading rates. Two age models are developed that combine (a) OSL and post-IR IRSL ages and (b) OSL ages and ‘expected’ ages from tying unit boundaries to the marine record. If the luminescence model is regarded as accurate, differences between this and the OSL/marine age model raise questions over the accuracy of the latter, as well as the processes controlling the zeroing of luminescence dates. The luminescence based age model is then used to derive the first fully independent reconstruction of climate proxies and accumulation rates from Carpathian loess. Such reconstructions can be used to compare to other independent records without assumptions inherent in correlation-based approaches. The findings demonstrate how variable accumulation rate is at the site, and compared to other independently dated Carpathian loess records. Average values vary north–south but are of similar order throughout the basin. Accumulation rate was highest during the later part of the last glacial, but variation on millennial timescales does not always match shifts in grain-size, suggesting diverse and complex influences. Environmental reconstructions using grain-size and magnetic susceptibility show that no one atmospheric system or air mass can explain the changes in the Carpathian Basin and that millennial-scale variability can only intermittently be tied to North Atlantic Heinrich events. Expanded ice sheets during the peak last glacial, combined with other atmospheric teleconnections, may have served to develop a strong anticyclone in the region. It was likely windier during earlier parts of the last glacial, but Atlantic and Mediterranean moisture was probably less abundant than during more humid interglacials.