Sorghum [Sorghum bicolor (L.) Moench] is one of the most important cereal crops grown in arid and semi-arid regions of the world. The North Eastern regions of Ethiopia are known for its high sorghum production and genetic diversity, and proneness to moisture stress. Globally, moisture stress is one of the major sorghum production constraints limiting genetic gain through breeding. Although, the importance of Ethiopia’s sorghum germpalsm has been widely recognized both nationally and internationally, the genetic potential of the germplasm has not yet been fully assessed and exploited in breeding programmes. Therefore, the objectives of this study were: (1) to evaluate sorghum production systems and patterns, major production constraints and cropping mechanisms, varietal diversification, farmers’ criteria for choosing varieties over time and space, and adoption of improved varieties, (2) to assess the agro-morphological and molecular diversity and population structure of lowland sorghum landraces collected from different geographic origins using morphological and SSR markers, (3) to assess the performance of sorghum landraces under moisture stress conditions and identify promising lines, and (4) to determine heterosis and combining ability of lowland sorghum landraces for grain yield, yield components and drought tolerance and to identify suitable parents for future hybrid cultivar development for the North Eastern Ethiopia or similar environments. A survey was conducted in the North Eastern Ethiopia sampling three Administrative Zones, six Districts and 12 Peasant Associations. Data was gathered from a total of 171 farmers and analyzed using SPSS statistical package. The results suggest that the performance of sorghum was generally poor mainly due to moisture stress, pests, diseases, weeds, farmland fragmentation due to demographic pressure, poor soil fertility, and poor performance of the local varieties. The productivity of sorghum was also largely hindered by the use of inherently poor yielding local landraces as farmers were forced to abandon their high yielding, big-headed and late maturing sorghum varieties due to the prevalence of recurrent moisture stress. The survey found that the region is as a rich source of genetic diversity and more than 70 distinct sorghum landraces were identified. The majority of the farmers grew their local landraces, despite the accessibility and availability of many improved sorghum varieties and production packages. Farmers were willing to adopt the improved varieties if they had farmers preferred attributes such as as red seed colour, tall with high biomass yield. To benefit the most from the available improved technologies, farmers have to be part of the breeding process right from the very beginning. Lowland sorghum accessions which exhibited farmer-desired traits were selected from the entire landrace germplasm collection at the national gene bank of Ethiopia. Field evaluations of the selected 278 landraces together with checks were held at Sirinka and Kobo agricultural research stations for 12 qualitative and 10 quantitative traits under stress and non-stress conditions. Two hundrad landraces were selected on the basis of their morphological distinctiveness and drought tolerance, in terms of earliness and yield stability. Molecular level diversity assessment was conducted using 30 SSR markers. Considerable magnitude of variation was observed among landraces between and within geographic origin for most of the traits studied. The morphological variability was also complemented by high molecular markers diversity. Thirty two pure lines were selected for inclusion as parents in the sorghum breeding programme for yield and drought tolerance. The selected lines were then crossed to four cytoplasmic male-sterile lines that had different cytoplasm systems (A1 to A4) using a line x tester mating design scheme. The 32 parents, together with the 128 hybrids and 4 check varieties were evaluated for grain yield, yield components traits and drought tolerance under stress and non-stress environments. Data were analysed using GenStat statistical package following a fixed effects model. Non-additive gene action was predominant in controlling plant height, grain yield, above ground biomass, grain filling duration, 100-seed weight and panicle weight, whereas additive gene action was found more important in controlling days to 50% anthesis and panicle length. Novel landraces with high GCA effects were selected including 214838-A, 242039-B, 75454, 73056-B, and 242050-A which will serve as potential parents for cultivar development. Similarly, the study identified new experimental hybrids i.e. ICSA 749 x 242039-B, ICSA 756 x 242049-B, ICSA 756 x 75454, ICSA 756 x 73059 and ICSA 756 x 214855 with high SCA effects and heterosis for grain yield which will be forwarded for further stability analysis and farmers participatory selections at representative growing environments. In general, the study identified invaluable sorghum germplasm and candidate hybrids useful for further breeding and conservation strategies. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2012.