The majority of farmers in Tanzania have not yet adopted modern maize varieties and still cultivate landraces and open pollinated varieties (OPVs) with low production potential and susceptible to diseases like maize streak virus (MSV), grey leaf spot (GLS) and northern corn leaf blight (NLB). The NLB disease is among the major causes of low yield and has been reported in all 21 maize growing regions in Tanzania. Breeding for host plant resistance with high yielding potential and involving the community in the breeding process is expected to address the problem of low yield, NLB disease susceptibility and low rate of F1 hybrid adoption. Therefore, the study was conducted to obtain additional sources of resistance to NLB disease, high yielding cultivars with community acceptable traits adapted to Tanzanian conditions. The main objective was to contribute to increased maize productivity in the western zone of Tanzania. The specific objectives of this study were therefore to : 1) investigate maize production limiting factors for smallholder farmers in western Tanzania, 2) identify farmers and stockist perceptions, opinions and maize variety selection criteria in western Tanzania, 3) establish NLB disease status in farmers’ fields of western Tanzania, 4) determine the genetic relationships among landraces and assess maize landraces as sources of breeding materials, 5) determine the combining ability and heterosis for NLB disease resistance of eleven maize inbred lines adapted to Tanzanian conditions, and 6) determine the gene action and inheritance of resistance to NLB disease in five maize inbred lines adapted to Tanzanian conditions. The study was conducted from 2008-2011 in three diverse environments which represent all the maize growing regions in the country The participatory rural appraisal (PRA) was conducted in three districts to investigate farmers’ and stockists preferred traits for maize selection in western Tanzania, determine maize production constraints facing farmers and assess NLB disease prevalence in the same area. A focus group of 30 farmers was selected in each of the three villages. Transect walks, wealth ranking and historical profiles were used in an informal survey. One hundred and fifty questionnaires were used in a formal survey. The recorded yield was only 1 t haˉ¹. Thirteen major maize production constraints, 13 insect pests and vermin and, 11 diseases were recorded. The NLB disease was reported to be increasing in severity in all farmers’ fields. Farmers’ preferred traits included resistance to abiotic and biotic stresses, early maturity, preferred milling qualities, high storage qualities and high yielding potential. Stockists mentioned 12 preferred maize variety traits which included high yielding, disease and insect pest resistance, heavy grain, large cob size and large grain sizes. Similarity between farmers and stockist variety preference ranking were found to exist. The occurrence and distribution of northern leaf blight (NLB) disease study was conducted to assess the incidence and severity of NLB disease in farmers’ fields in seven districts. The study was conducted for two seasons. In each season, 175 fields with 5600 plants were sampled. There were sixteen varieties grown with wide NLB disease reaction variation. Gembe, a landrace, was among the three observed resistant varieties. The NLB disease has changed its distribution pattern affecting all districts of the western zone. The disease incidence in season two (2009/2010) significantly increased from season one (2008/2009) t= -3.25 (348), P= 0.001. About 30% of both means of blight incidence and severity were recorded in the area. Characterization and screening of maize landraces for northern leaf blight disease resistance was conducted to determine the genetic relationships among landraces, assess maize landraces as sources of NLB disease resistance and assess important agronomic traits for future maize improvement. Ninety breeding materials consisting of 71 landraces and 19 commercial varieties were evaluated. The average yield of landraces under research management was 2.3 t haˉ¹. Landrace TZA 3075 was identified as NLB disease resistant. Yield potential, dent grain texture, white endosperm and husk cover were important agronomic traits observed among landraces. There were high variations in terms of morphology and NLB disease resistance among the landraces. Five principal components contributed to 71.98 % of total variation. Clusters analysis revealed five distinct groups of landraces. Leaves/plant, infested leaves/plant, lesion number, lesion length, lesion width and NLB disease incidence traits highly contributed to variation and grouping of landraces. Combining ability analysis for northern leaf blight disease resistance was conducted to estimate the combining ability for NLB disease resistance of 11 maize inbred lines adapted to Tanzanian conditions, determine maternal effects which are involved in NLB disease resistance in maize germplasm, and determine the heterosis in the F1 hybrids. A full 11 x 11 diallel cross was performed. All top ten experimental hybrids in each of the three sites had negative midparent heterosis for NLB disease severity. The overall mid-parent heterosis means for yield across sites was 152%. The mean sum of squares for GCA was highly significant (P< 0.001) on disease severity indicating additive gene action effects. Mean sum of squares for SCA were highly significant for disease severity and yield implying non-additive gene action effects. The mean squares for reciprocal effects were highly significant on yield and non-maternal sum of squares had significant effect (P<0.05) on yield. The GCA contribution was high for disease severity (91%) and lesion number (85%). Almost, all GCA effects for NLB disease resistance were negative implying contribution to disease resistance. Due to preponderance of the additive gene action, recurrent selection could be used to improve the resistance of inbred lines while the non-additive gene action could be exploited in breeding for disease resistant hybrids. Generation mean analysis of northern leaf blight disease resistance was conducted to determine the mode of gene action involved in the inheritance of resistance to NLB disease in five inbred lines adapted to Tanzania at contrasting environments, estimate heterosis and heritability in five tropical inbred lines. Generation mean analysis was conducted using a six parameter model comprising P1, P2, F1, F2, BCP1 and BCP2 generation progenies. The mean sum of squares for environment, replication with the nested environment, generations, generations x environment interactions were highly significant (P<0.001). The full model of additive, dominance, additive x additive and additive x dominance epistatic effects was highly significant (P<0.001). Nonetheless, the additive gene effects were predominant ranging between 57% and 89% which was matched by large heritability (54%-85%). The average degree of dominance ranged between -0.52 and 0.88 supporting observations of partial dominance. The NLB disease severity showed a continuous distribution in all three sets for F2, BCP1 and BCP2 populations which is an indication of quantitative nature of inheritance and additive gene effects. The mid parent heterosis ranged from -19 to 1%. Therefore, resistance to NLB disease could be improved through selection by exploiting the additive gene effects. The epistatic gene effects would cause less complications because they were negligible (<25%). The client oriented breeding for maize northern leaf blight disease resistance was carried out to perform farmers and stockists assessment on the 110 F1 experimental maize hybrids and compare them with breeders selection criteria. Breeders selection criteria ranked 10 top high yielding experimental hybrids. Farmers developed 14 while stockists developed 13 selection criteria. The most preferred hybrids by farmers were VL 05616 x CML 159, CML 159 x KS03- 0B15-47 and EB04-0A01-304 x CML 442 while stockists preferred VL 05616 x CML 395, EB04-0A01-304 x CML 442 and VL 05616 x CML 159. Two F1 experimental hybrids EB04- 0A01-304 x CML 442 and CML 159 x CML 442 appeared in all top five ranked hybrids by breeders, farmers and stockists. Generally, findings showed that, farmers, stockists and breeders coincide in some selection criteria but also differ in other cases. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2012.