Finger millet (Eleusine coracana (L.) Gaertn. ssp. coracana) is an important food, food security and cash crop in eastern and southern Africa where small-scale farmers grow it in low input farming systems. The crop has food security, nutritional, cultural, medicinal, and economic value with high industrial potential. Little research and hardly any breeding have been done on the crop leading to low yields and low production. A project was therefore implemented in western Kenya during 2004-2007 seasons to investigate the possible breeding contributions to enhance productivity and production of the crop. The research comprised a social survey, germplasm evaluation, appraisal of ethrel as a chemical hybridising agent (CHA), genetic analysis of yield, and resistance to blast and Striga, and breeding progress in developing new finger millet varieties. A participatory rural appraisal (PRA) was conducted in three districts during 2006 to position finger millet (FM) in the farming systems, production constraints, and variety diversity and farmer preferences. The PRA established the high rating the peasant farmers gave to finger millet among crop enterprises, using it for food, cash, brewing, ceremonies and medicinal purposes. Farmers cultivated many varieties ranging from five to nine in a district, but each district had its own popular variety. Farmers used the following criteria to select new cultivars: high yield potential; early maturity; resistance to blast disease, Striga, birds, drought, and lodging; large head size, dark grain colour, and good taste. This probably indicated the willingness of farmers to adopt new varieties. Farmers identified constraints to production as blast disease, Striga, wild FM, birds, rats, termites, lack of market, labour shortage, and low yield. The farmers’ variety selection criteria and production constraints underscored the need to improve finger millet varieties. Evaluation of 310 accessions for trait variability and association conducted during 2005 long rain (LR) season at two sites revealed wide variation among the accessions for yield and secondary traits. The best accessions grain yield was above the yield potential of 5,000- 6,000kg ha-1 reported in other environments. Accessions KNE 072 (7,833kg ha-1), GBK 028463 (7,085kg ha-1), GBK 029661 (6,666kg ha-1) and FMBT ACC#42 (6,566kg ha-1) were outstanding. The data showed the opportunity to select for yield directly because of its wide variability but indirect selection could also be used to exploit seedling vigour as shown by its high correlation to yield and direct and indirect positive effects on yield through plant height and single plant yield in path analysis. The wide genetic variability among the genotypes for several traits indicated high potential to breed new and better finger millet varieties. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2008.