Soil moisture is the vital component of the hydrological cycle and its variability is largely uncertain in the upcoming decades. In this paper the future projections of soil moisture changes over South Asia have been analyzed on both annual and seasonal basis from 2020-2050. The comparison of 24 CMIP5 (Coupled Model Inter comparison Project Phase 5) models with GLDAS (Global Lad Data Assimilation System) reanalysis has been done to assess their performance in simulation of soil moisture (0.1m) over the South Asian region. The BCC-CSM-1, MIROC5 were the two models that captured the soil moisture conditions well. They were selected on the basis of high correlation significant at 95 %, least RMSE and Standard Deviation. The mean of both the models were acquired to view the fidelity in the surface layer soil moisture changes in the South Asia. Under RCP’s 4.5 and 8.5 BCC-CSM-1 and mean mostly showed slight to moderate decrease in soil moisture content both annually and seasonally (DJF, MAM, JJA and SON). While MIROC 5 projected slight to extreme increase in soil moisture content in the northern half of the South Asia and slight to extreme decrease over the southern and eastern parts of the South Asia annually and season-wise (DJF, MAM, JJA and SON). The MAM will be having the lowest moisture content in the region from 2020-2050. It is anticipated that if soil moisture stress projected by the two models and their mean becomes persistent in future, it may lead to the agriculture drought and generate food security threat in the South Asian region.