Bananas and plantains (Musa spp.) are cultivated commercially as a dessert fruit and by small-scale farmers as a staple food crop in many countries of the world. Sustainable production, however, is threatened by a number of diseases and pests such as Fusarium wilt (Fusarium oxysporum f.sp. cubense), black Sigatoka (Mycosphaerella fijiensis), the banana weevil (Cosmopolitus sordidus) and the burrowing nematode (Radopholus similis). Many diseases and pests of banana cannot be managed by conventional control methods, and replacement cultivars are often not acceptable to local industries. Resistance can be introduced into banana by means of conventional and unconventional improvement methods. Conventional breeding programs have many limitations, due to sterility of cultivated bananas, long growth cycles, low seed set and hybrids that are often not accepted by consumers. Unconventional improvement for enhanced resistance involves methods such as in vitro mutagenesis, protoplast culture, and genetic modification. During genetic modification, foreign genes are introduced into banana by means of Agrobacterium-mediated transformation or by particle bombardment. One of the most powerful means to reduce the impact of pests and diseases is the use of somatic embryogenesis for unconventional plant improvement and the propagation of disease-free plants. In this thesis, immature male flowers of Grande Naine (Musa acuminata, Cavendish subgroup, AAA) were isolated and incubated on MA1 medium to form somatic embryos with ideal callus. When ideal callus was transferred to liquid MA2 medium, a heterogeneous cell suspension was formed. Non-embryogenic aggregates were removed to ensure a cell suspension constituted of small embryogenic clusters only. Somatic embryos were obtained from the cell suspension after plating the embryogenic clusters on solid MA3 medium. These somatic embryos were transferred to MA4 medium for germination, and to P6 medium to develop into in vitro plantlets. Embryogenic cell suspension can be used for genetic engineering of disease and pest resistant plants, in vitro mutagenesis, germplasm conservation and protoplast culture. An Agrobacterium-mediated transformation system was established for the improvement of Cavendish banana cultivars, the only bananas produced for the fresh fruit market in South Africa. Embryogenic cell suspensions from the cultivar Grande Naine were co-cultivated with Agrobacterium strains harbouring the plasmids pCambia1305.1, pART-TEST7 and pKYΩOCI. Antibiotic-resistant embryos derived from transgenic cell suspensions developed into banana plantlets 12 weeks after cultivation on MA4 medium. In total, 145 putative transgenic plants were produced. Molecular analysis revealed that the Gus gene was integrated into the genome of transformed plants, and a histochemical GUS assay showed that the Gus gene was expressed in putative transgenic plants. In future, southern blot assays will be performed to determine copy numbers of the transgenes, and the putative transgenic plants containing the OcI gene tested in the greenhouse and the field for resistance to the banana weevil and the burrowing nematode. The successful transformation of Grande Naine reported in this work will contribute significantly towards improving Cavendish bananas in South Africa, and offers the opportunity to modify banana and plantain varieties cultivated in Africa for enhanced disease and pest resistance. A highly discriminative fingerprinting technique, called amplified fragment length polymorphisms (AFLPs), was used to differentiate between closely related species within the Cavendish banana subgroup. AFLP profiles of eight banana varieties, cultivars and hybrids were generated on a Licor analyser using seven different primer combinations. Results showed that the banana plants were subdivided in clades according to their genomic composition. More importantly, the AFLP technique was able to separate the different cultivars within the Cavendish subgroup. Hopefully this technique could eventually be applied to accurately detect somaclonal variants in banana due to micropropagation and genetic modification.