Despite much effort over the last decade, there still remain many uncertainties in the assessed impacts of climate change on water resources. This study has carried out Monte Carlo Simulations to characterise the sampling uncertainties in assessed water resources impacts. The investigation employed data from catchments in northeast England, which incorporate water supply reservoirs. The impacts assessment used scenarios from three GCM experiments: (i) the Canadian first generation coupled model (CGCM1), (ii) the Australian first generation coupled model (CSIRO-mk2b) and (iii) the British third generation model (HadCM3). The results showed that yield impacts are subject to wide variability, irrespective of the GCM experiment, which calls for caution when using mean impacts obtained from single data record analysis for decision making. © 2006 Springer Science+Business Media, Inc.