Infonnation about the amount and distribution of genetic variation in gennplasm collections isrnimportant for efficient management of gennplasm collections and for effective utilization ofrnsuch materials in plant breeding. Sorghum (Sorghum bie%r (L.) Moench) is one of the mostrnimportant cereals worldwide and its tolerance to drought makes it particularly important inrnsemiarid tropical regions, including Ethiopia and Eritrea. The significance of sorghumrngennplasm from these countries for worldwide improvement of sorghum is also wellrnacknowledged. However, there is limited infonnation about the amount and distribution ofrngenetic variation in sorghum gennplasm from these countries. In this thesis, the amount andrnpattems of distribution of genetic variation was detennined in cultivated sorghum (s. bie%rrnssp. bie%r) gennplasm collections from Ethiopia and Eritrea using morphological traits,rnallozymes and random amplified polymorphic DNA (RAPD) markers. A similar investigationrnwas made on wild sorghum (s. bie%r ssp. vertieillijlorum) from Ethiopia using RAPDrnmarkers.rnThe results show that the materials exhibit a wide range of variation for quantitativerntraits and a high Shannon-Weaver diversity index (il= 0.90) for 10 qualitative traits. Inrncontrast, the level of allozyme variation was found to be low (mean expected heterozygosity =rn0.024). The level of RAPD variation was found to be intermediate both in cultivated (il=0.53) and wild sorghum (H= 0.49). The lack of correspondence among the three levels ofrnvariation revealed by these methods points at the difficulty of making prediction of variationrnof one based on the other and explanations are provided for the possible causes of therndisparity. Despite the disparity in the levels of variation from the three data sources (i.e. ,rnmorphological traits, allozymes and RAPD markers), the structure of the variation was fOlmdrnto be similar for all the data sources. In all, the within accessions, regions and the withinrnadaptation zones variation accolmted for a large portion of the total variation compared withrnthe between accessions, regions and the between adaptation zones variation. Based on thesernresults, sampling more accessions within each regions and adaptation wnes and morernindividuals within populations are suggested as future sampling strategies. Grouping of thernmaterials into lowland, intermediate and highland elevations seems to be justified only forrndiscriminant analysis of quantitative traits, otherwise, there was no convincing evidence thatrnsuch materials are genetically substructured. Regions with overall high diversity are identifiedrnand suggested for in situ conservation. The results from quantitative trails indicate thatrnaccessions from Eritrea could be important sources of genes for early maturity. Detailedrnimplications of the results for germplasm conservation and utilization are discussed and futurernlines of research are suggested.rnKey words: alJozymes; genetic variation; germplasm; morphological variation; RAPD; Sorghum bicolor