Thirty-seven tomato genotypes of different geographical origin were assessed ini a field experiment conducted in Raipur, Madhya Pradesh, India, during 1998 to determine the nature and magnitude of genetic divergence using a non-hierarchical clustering approach with the help of Mahalnobis D² statistics for yield and its contributing characters. The population was grouped in four clusters. The clustering pattern indicates that there was no association between geographical distribution of genotypes and genetic divergence. The genetic drift and selection in different environments can produce greater diversity than geographical distance. The characters, namely, number of primary branches, longitudinal fruit length, days to flowering, pericarp thickness, plant height and average fruit weight contributed to maximum divergence and played a more major role in the improvement of tomato yield.