A DESSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE AWARD OF MASTER OF SCIENCE IN CROP SCIENCE OF SOKOINE UNIVERSITY OF AGRICULTURE, MOROGORO, TANZANIA.One hundred maize genotypes of different categories were evaluated for Maize Lethal Necrosis Disease (MLND) resistance in three locations under natural infestation. Sixty inbred lines, thirty landraces and ten improved varieties were subjected to disease hot spot areas. Experiment was conducted at Ngaramtoni in Arusha municipality, Mlangarini in Arumeru District and Kiru six in Babati Rural Distriduring 2014 and 2015 seasons. The trial was laid down in Randomized Incomplete Alpha Lattice design and replicated three times. Breeding nursery was established in an un-replicated trial at Kirusix in 2014 off season. Single cross hybrids were developed using 6x6 full diallel fashion following Griffing’s (1956) design I Model I. The parental materials used were drawn from diverse inbred lines. Evaluation trials were conducted in three locations (Ngaramtoni, Mlangarini and Kirusix). The trials were laid down in a Randomized Complete Block Design with three replications. All genotypes were evaluated for resistance against MLND and yield components. Analysis of variance showed significant differences among treatments at (p≤0.001). No genotype showed complete immunity against MLND. However, landraces showed some resistance scoring from 3-4 compared to inbred lines and improvedvarieties where in most of the locations they scored4-5.Nature of gene action and genetic parameters for disease resistance were studied in a diallel cross involving six maize inbreds. The adequacy of genetic model was determined through regression coefficient and covariance – variance (Wr-Vr) test to validate the data set. The data were analysed according to Hayman’s analysis of variance and components of genetic variance were estimated. Additive genetic effects appeared to be more pronounced in the genetic control than non-additive. The parent CML 144 was found to be the best combiner with GCA of (-0.556***) while CML 503 and CML 444 were found to be among poor combiners with GCA of 0.62*** and 0.231**respectively. The graphic analysis revealed thatallelic distributions were highly influenced by environment for some genotypes. Since there is high genetic variation among the genotypes studied, selection for the promising material can be done successfully.