ABSTRACT A probabilistic framework for evaluating changes in offshore wind potential under a changing climate is presented considering three general circulation models ( GCMs), two representative concentration pathways and two statistical downscaling techniques. The analysis was done with respect to three specific locations along the Indian coastline, identified earlier for future offshore energy extraction. The uncertainty among the GCMs was addressed and the choice of an appropriate GCM was made by combining analysis of GCM uncertainty and model reliability. The regional scale wind was derived from the selected GCM using one hard computing method, namely bilinear interpolation and quantile mapping, and one soft method, an artificial neural network. The efficiency of these methods was assessed using certain performance criteria for the historical period and a convergence criterion over the future period. To evaluate wind potential, the GCM output was downscaled from daily to hourly, taking into account the removal of gusts because of long interval averaging. The wind potential for past (1979-2005) and future (2006-2032) time slices was determined using both hard and soft methods as well as two representative concentration pathways. The resulting four outcomes at annual and seasonal scale were compared with the extrapolated trend of past wind potential derived from both GCM and reanalysis data. Finally, the most efficient strategy was evaluated and the actual extractable power was estimated considering a standard power curve. It is concluded that at all the three locations the annual average wind potential will substantially increase, benefitting the offshore industry. [ABSTRACT FROM AUTHOR]