The first comparative and systematic climate model study of the sensitivity of the climate response under Last Glacial Maximum (LGM) conditions to freshwater perturbations at various locations that are known to have received significant amounts of freshwater during the LGM (21 kyr BP) climate conditions is presented. A series of ten regions representative of those receiving most of the meltwater from decaying ice-sheets during the deglaciation is defined, comprising the border of LGM ice-sheets, outlets of rivers draining part of the melting ice-sheets and iceberg melt zones. The effect of several given freshwater fluxes applied separately in each of these regions on regional and global climate is subsequently tested. The climate response is then analysed both for the atmosphere and oceans. Amongst the regions defined, it is found that the area close by and dynamically upstream to the main deep water formation zone in the North Atlantic are most sensitive to freshwater pulses, as is expected. However, some important differences between Arctic freshwater forcing and Nordic Seas forcing are found, the former having a longer term response linked to sea-ice formation and advection whereas the latter exhibits more direct influence of direct freshening of the deep water formation sites. Combining the common surface temperature response for each respective zone, we fingerprint the particular surface temperature response obtained by adding freshwater in a particular location. This is done to examine if a surface climate response can be used to determine the origin of a meltwater flux, which is relevant for the interpretation of proxy data. We show that it is indeed possible to generally classify the fingerprints by their origin in terms of sea-ice modification and modification of deep-water formation. Whilst the latter is not an unambiguous characterization of each zone, it nonetheless provides important clues on the physical mechanisms at work. In particular, it is shown that in order to obtain a consistent see-saw temperature pattern, addition of freshwater in the Northern Hemisphere at sites dynamically close to the deep water formation zones is needed. Finally a preliminary data-model comparison for the time of the Heinrich event 1 suggests that those sites are indeed the most favourable to explain the pattern of climate variability recorded in proxy data for this period. More importantly, this model-data comparison enables us to clearly reject a substantial fraction of the zones tested as potential source for large freshwater entering the ocean at that time. © 2009 The Author(s).