Soil contaminant deposition is highly dependent on anthropogenic activities while contaminant retention, mobility and availability are highly dependent on soil properties. The knowledge of partitioning between soil solid and solution phases is necessary to estimate whether deposited amounts of contaminants will rather be leached through runoff or accumulated. Besides pedological driven partitioning, runoff is expected to change during the next century due to changes in climate and in rainfall patterns. In this study, we aimed at estimating at the European scale the areas concerned by potential risk due to contaminant leaching (LP). We also defined in the same way the surface areas where limited Cu leaching occurred, leading to potential accumulation (AP) areas. Among contaminant, we focused on copper (Cu) widely used in agriculture, resulting in high spatial variations in deposited and incorporated amounts in soils. We developed a method using both Cu partition coefficients (Kf) between total and dissolved Cu forms, and runoff simulation results for historical and future climates. The calculation of Kf with pedo-transfer functions allowed us to avoid any uncertainties due to past management or future depositions that may affect total Cu concentrations. Areas with high potential risk of leaching or of accumulation were estimated over the XXIth century by comparing Kf and runoff to their respective European median. Thus, at three distinct times, we considered a grid point at risk of LP if its Kf was low compared to the European median and its runoff was high compared to the European median of the time. Similarly, a grid point was considered at risk of AP if its Kf was high and its runoff was low compared to their respective European median of the time. To deal with uncertainties in climate change scenarios and the associated model projections, we performed our study with two representative atmospheric greenhouse gases concentration pathways, defined with climate change associated to a large set of socio-economic scenarios found in the literature. We used two land surface models (ORCHIDEE and LPJmL, given soil hydrologic properties) and two global circulation models (ESM2m and CM5a, given rainfall forecast). Our results show that, for historical scenario 6.4 ± 0.1 % (median,median deviation) and 6.7 ± 1.1 % of the grid cells of the European land surfaces are concerned by LP and AP respectively. Interestingly, our results simulate a constant global surface concerned by LP and AP, around 13% of the grid cells, consistent with an increase in AP and a decrease in LP. Despite large variations in LP and AP extents depending on the land surface model used for estimations, the two trends were more pronounced with RCP 6.0 than with RCP 2.6, highlighting the global risk of combined climate change and contamination and the need for more local assessment. Results are discussed to highlight the points requiring improvement to refine predictions.