Root architecture of 220 diverse barley germplasm of Indian (134) and exotic (86) origin was evaluated for polyethylene-glycol simulated drought stress to identify drought tolerant genotypes. The evaluation of root images was done using root scanner (WinRHIZO Pro software v2009). Variance, Principal Component Analysis (PCA) and Ward's agglomerative hierarchical clustering were carried out using SAS software. Correlation matrices were generated using R. Analysis of variance indicated that under stress treatment, differences among the tested germplasm accessions were highly significant (P) with respect 0.01 to total root length (RL), seminal root number (SRN), root surface area (RSA), root volume (RV), root diameter (RD), lateral root number (LRN) and root dry weight (RDW) per seedling. LRN was stimulated while other root traits such as RL, RSA, RV and RDW were significantly inhibited under stress. PCA indicated that first three components accounted for 80.50% of the total multivariate variation with PC1 accounting for 44.83%, PC2 for 19.84% and PC3 for 15.87% and it was mainly explained by RL, RSA, RV and RDW. Cluster analysis grouped 220 barley accessions into five major clusters, with cluster I being drought susceptible, cluster II being drought tolerant, cluster III being moderately drought tolerant, Cluster IV being highly drought tolerant and Cluster V being highly drought sensitive, respectively. Accessions IC393980, IC082719, IC329556, EC492318, EC578789, EC578790, IC335811 and wild barley H. marinum ssp. gussoneanum proved to be potential genetic resources for drought tolerance, which can be used in cereal breeding program for rain-fed agriculture.