Understanding climate variability in the past, the present and the future is currently the focus of many research projects world-wide. For the most part, such work aims to provide best estimates for climate using (1) measured instrumental data about modern climate and its impacts, (2) available historical and proxy data about past climate and (3) our current understanding of the climate system as implied by climate models of varying degrees of complexity. With increased interest in comparing and combining climate estimates based on these different sources, for different places in time and space, it seems that understanding uncertainty (in proxies, models and observations) may be as important as the actual estimates of climate themselves. This session encourages researchers working on climate reconstruction, observation and prediction to think and talk about understanding, quantifying and modelling the various sources of uncertainty that have an impact on their work. Although the session is proposed by members of an international network known as SUPRAnet whose focus is to study uncertainty in palaeoclimate reconstruction, we encourage submission of papers relating to uncertainty on any aspects of work relating to reconstructions, estimates or predictions of climate.