The type and level of genetic variance in one hundred and fourteen F5 families of rice obtained from six different crosses was estimated along with their seven parents using Mahalanobis D 2 -statistics by considering 10 characters. Mahalanobis D 2 analysis revealed considerable amount of diversity in the material. The genotypes were grouped into twelve clusters. Cluster IX constituted maximum number of genotypes (26). The genotypes falling in cluster VII had the maximum divergence, which was closely followed by cluster VIII and cluster XI. The maximum inter cluster D2 values was observed between cluster X and XI (931.276) followed by cluster VIII and XI (814.784) suggesting that the genotypes constituted in these clusters may be used as parents for future hybridization programs. Principal component analysis revealed that families MTU 2462-1-5-2, MTU 2462-15-1-1, MTU 2468-30-2-2, MTU 2468-29-3-1, MTU 2462-1-9-2, MTU 2469-23-2-1, MTU 2469-6-2-1, MTU 2469-36-1-1, MTU 2469-32-2-1 were more divergent and hence, can be used in breeding programs. Hence, the results of cluster analysis were supported by principal component analysis.