Thirty okra genotypes of diverse eco-geographical origin were grown in single-row plots in a randomised complete block design. The data collected on 14 characters were subjected to analysis of variance. By multivariate analysis (Mahalanobis D(2) technique), the genetic divergence among the genotypes were quantitatively measured. The genotypes were grouped into five clusters by this technique. There was no relationship between clustering pattern and eco-geographic distribution. The effects of genetic divergence on the choice of parental stock in hybridization was discussed.