The soybean programme in Zimbabwe is over seventy years old. However, there is lack of information on breeding gains, genetic diversity, heritability, genetic advance, combining ability, gene action and relationships between grain yield and secondary traits available for breeding. Therefore, the aim of the present study was to characterise the genetic diversity of the available germplasm, determine gene action conditioning grain yield and estimate the breeding gains that have been realised since the inception of the breeding programme. Evaluation of 42 soybean genotypes for genetic diversity conducted during 2010/11 and 2011/12 cropping seasons, using phenotypic and molecular characterisation approaches, revealed evidence of wide diversity among the genotypes. The phenotypic traits and SSR markers assigned the soybean genotypes to 8 and 15 clusters respectively. The SSR marker technique was more polymorphic, informative and highly discriminatory. The clustering pattern and relatedness from SSR data was in agreement with the pedigree data while the phenotypic clustering was divorced from pedigree data. Genotypes, G41 and G7; G41 and G1; G41 and G42 were the most divergent; therefore, they could be utilized as source germplasm in cultivar development and commercial cultivars. Investigations on breeding gains involving 42 cultivars (representing a collection of all the varieties that were released in Zimbabwe from 1940 to 2013) showed that improvement in grain yield was slowing down. However, annual genetic gain was estimated to be 47 kg ha-1 year-1 representing an annual gain of 1.67%. Furthermore, grain yield ranged from 2785 to 5020 kg ha-1. Genotypes, G16, G15, G17, G1 and G42 exhibited superior performance in grain yield and other agronomic traits and are therefore, recommended for utilisation in the hybridisation programme. Seed protein concentration decreased by 0.02 year-1 while oil increased by 0.02, 100 seed weight increased by 0.21 g year-1 over time. In addition, number of days to 95% pod maturity and pod shattering increased by 0.35 and 0.38 days year-1 respectively while lodging declined by 0.31%. Results indicated that emphasis should be refocused on grain yield to restore the original linear increase. Assessment of the magnitude of GEI and stability of 42 released cultivars was done over 13 environments and two seasons using additive main effects and multiplicative interaction, cultivar superiority and rank analyses. Results showed that environment and GEI captured larger portion of the total sum of squares, which reveals the influence of the two factors on grain yield, hence, the need for evaluating soybean genotypes in multi-environment trials and over years. Further, the data revealed that GEI was of a crossover type because of differential yield ranking of genotypes. The three stability parameters selected two genotypes, G1 and G15, as the most productive, consistent and stable, thus they could be produced in diverse environments while G2, G4, G5, G7, G16, G40, G17, G18 and G31 were identified as unstable and suitable for specific adaptation. Correlation and path analyses showed that grain yield was positively and significantly correlated with number of branches per plant, number of nodes per plant, shelling percentage, and number of days from 95% pod maturity to first pod shattering, implying that breeding and selection for these traits probably improved grain yield. Number of nodes per plant, plant height and 100 seed weight exhibited highest direct effects on grain yield while, number of nodes per plant and plant height presented the highest indirect effects on grain yield. These results demonstrated that number of nodes per plant and plant height could be recommended as reliable selection traits for developing high yielding genotypes of soybean. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2013.