Sorghum breeders have not made much yield improvement in new sorghum varieties in Kenya since Serena in was released in the late 1960s. KARI Mtama-1 which was released in 1993 has no yield advantage over Serena. A yield plateau for sorghum in Kenya is apparent. A new breeding approach was adopted to break that yield barrier. Development of hybrid sorghum was proposed and is expected to break the yield barrier and also deliver cultivars that meet farmers' main requirements. The objectives of the study were to (1) identify farmers' requirements in sorghum cultivars, constraints to sorghum production and why improved cultivars from research are not being adopted, (2) characterize male and female parents and establish if genetic distance could identify superior parent populations for hybrid production (3) estimate genetic variance components and determine the possibility of using GCA and SCA estimates in choosing parents for use in hybrid production, (4) test hybrids and open pollinated variety (OPV) parental lines for stress tolerance and identify tolerant hybrids for further testing and, (5) compare single cross hybrids and OPV varieties in yield performance. Participatory rural appraisal in Kitengei and Nzambani areas of Kambu showed that sorghum was especially important in semi-arid parts of Kenya. Food, trade, feed, nursing food and thatching were the most important uses of sorghum. High grain and stover yield, large grain size, early maturity, drought tolerance, pest and disease resistance, coloured grain and intermediate plant height were the major requirements of farmers. Fifty-three pollinators and forty-one male sterile parents were introduced from four sources and screened together with 27 pollen parents from Kenya. Parents and hybrids were tested in 4 environments: high and low plant density, in high and low moisture regimes laid out in a triple square lattice design in Kenya, with parents having two additional tests in South Africa. Males, females, sexes and parental sources differed significantly in head weight. There were sex x country and sex x environment interactions for head weight. Genetically distant parents' populations had higher chances of superior heterosis. Parents showed significant additive genetic variance in head weight. The regression of non-additive to additive genetic variance was roughly one and significant. Three female and five male parents were suitable for production of hybrids adapted to multiple environments. Hybrids and OPV lines significantly varied in head weight. Hybrids were superior to OPV lines in most agronomic traits. Economic superiority of the hybrids was sufficient to cover cost of hybrid production and distribution in Kenya. Hybrids and OPV lines varied significantly for plant density stress. Hybrids were less sensitive to stress and more productive than OPV lines under population density stress. KARI varieties were sensitive to plant density stress. In general low sensitivity to stress was beneficial and hybrids had superior yield to inbred varieties. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2006.