In order to evaluate cold tolerance of twenty barley genotypes under field conditions an experiment was carried out with Randomized Complete Blocks Design at 3 sowing dates (October 6, November 5 and December 5) during 2011 - 2012 seasons. Also, another experiment was done on the same genotypes based on Completely Randomized Design with 5 replications under greenhouse conditions. Results analysis of variance showed that significant differences between sowing dates for traits of plant height, spike per m2, grain per spike, 1000 KW and grain yield. First sowing date at October had higher spike per m2, 1000 KW, grain yield and total dry matter. There were significant differences, for total dry matter and ion leakage at 4 and 6 leaves stage. The highest and lowest ion leakages were observed at Redut/OK84817 (37.30µds/m2) and Legia/CWB117-5-9-5 (21.44µds/m2) genotypes, respectively. Traits of 1000 KW, total dry matter had positive significant correlations with grain yield r = 0.62** and r =0.74**, respectively. It seems that these traits could be used an indirect criteria for cold tolerance evaluation. At regression analysis with stepwise method 1000-kernel weight and harvest index remained at final model. Cluster analysis with all traits showed that genotypes of 12, 16, 4, 3, 17, 8, 15, 7, 14, 20, 11 and 9 were superior genotypes. At principal component analysis four first components showed 80.9% from total variation. First component was important at improving grain yield and third components named as a grain yield components.
