Pigeon pea [Cajanus cajan (L.) Millspaugh] is a very important grain legume crop for food, cash and firewood in Malawi. However, its production is affected by Fusarium wilt (Fusarium udum Butler), which causes up to 100% yield loss. The deployment of resistant varieties would be an economical way to manage the disease, and for this, more information is needed on farmers' preferences for local landraces, how farmers and consumers can be involved in developing new varieties resistant to wilt disease, and the genetics of inheritance of resistance. This information would be used to devise a breeding strategy. A participatory rural appraisal was used in the southern region of Malawi to identify pigeonpea production and marketing constraints. Results showed that Fusarium wilt was the most prevalent and destructive disease of pigeonpea in the area. Other constraints included pests, flower abortion, low yields, and low soil fertility. Local landraces accounted for 84% of the pigeonpea production in Malawi. Local landraces were preferred due to their fast cooking time, taste, and the high prices they earn the farmer. Participatory variety selection was used to identify landraces with desirable traits that could be used in the breeding programme. Farmers and buyers selected ten local landraces which were used in the genetic improvement programme. Pigeonpea local landraces and international Crops Research Institute for the Semi-Arid Tropics (ICRISAT) genotypes were evaluated for wilt resistance, yield, and secondary traits at three sites over three seasons. Most of the landraces were susceptible to wilt and late maturing. However, AP10, a local landrace, was high yielding and resistant to wilt and outperformed ICRISAT varieties. This local landrace showed promise for use as a source material for Fusarium wilt resistance in other locally adapted farmer-preferred varieties lacking resistance. The local landraces needed genetic improvement in wilt resistance, yield, early maturity, number of branches and seeds pod. Laboratory and screenhouse studies were performed to develop a new Fusarium wilt screening technique. Grains of finger millet, sorghum, and wheat were tested as media for multiplying F. udum isolates. Pathogenicity tests were done on Bunda College and Bvumbwe Research Station isolates. The Bunda isolate was then used in an infested-seed inoculation technique against eight differential cultivars. The results showed that finger millet, sorghum and wheat were equally effective for rapid multiplication of F. udum isolates. Wheat grain showed the best results for pathogen multiplication and inoculation, due to the large seed size for easy handling. The inoculation process involved placing infested wheat grain on bruised pigeonpea roots and transplanting into soil in pots. The infested seed inoculation technique, which is the first of its kind for pigeonpea, was effective in screening pigeonpea for wilt resistance. The selected landraces were crossed with wilt resistant testers in a 12 lines x 4 testers mating scheme, and 48 F1 crosses were generated. These F1 crosses were evaluated for wilt resistance, yield, and secondary traits. The variations among F1 crosses for wilt and secondary traits were due to additive gene action in both parents and the dominance effects arising from the interactions of parents. Parental lines, with good combining ability effects for wilt resistance (AP2, AP3, and AP4), days to 50% flowering, seed pod, plant height, stem diameter, and number of primary and secondary branches were identified, while ICEAP00554 (tester) was a good general combiner for wilt resistance and days to 50% flowering. These lines would be useful in breeding for Fusarium wilt resistance in farmer-preferred pigeonpea genotypes in Malawi or similar environments. Specific F1 crosses were identified with significant SCAs for wilt resistance, days to 50% flowering, and secondary branches. The significance of GCA and SCA effects, which indicated importance of both additive and non-additive gene effects, respectively, suggested that both selection and hybridisation would be useful to improve the resistance in farmer-preferred varieties. Segregation analyses were conducted on F2 populations to determine the resistance to susceptibility phenotypic ratios. The Chi-square analyses showed that resistance to wilt was dominant over susceptibility in most F2 populations. The segregation ratios of 3:1, 13:3, 15:1, and 9:7 (R:S) indicated that either one dominant gene, or two inhibitory genes, or two independent dominant genes, or two complementary genes, respectively, were conferring wilt resistance in these crosses. Involvement of only a few genes governing wilt resistance suggested few complications, if any, in breeding for this trait in these locally adapted pigeonpeas. The Pedigree breeding method would be recommended for incorporating these traits. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2007.