Seventy four genotypes representing diverse geographic origin were studied for genetic divergence using Mahalanobis D2 statistic. These genotypes were grouped into twelve clusters. The mode of distribution of genotypes to various clusters was at random suggesting that there is no relationship between geographical distribution and genetic diversity. Based on inter-cluster distances, the clusters VII vs X, VI vs XII and X vs XII were found as divergent. Hence, selection of genotypes from these clusters namely ICGV 99032 (cluster VI), TCGS 647, JL 220 (cluster VII), ICGV 95477, JL 24, ICGV 99054, ICGV 86699 (cluster X) and ICGV 99029 (cluster XII) for hybridization programme may result into good recombinants. The characters 100-kernel weight, shelling percentage and harvest index contributed maximum towards genetic divergence in both D2 analysis and canonical root analysis. Further, canonical root analysis confirmed the clustering pattern obtained by D2 analysis.