The aim of this thesis was to develop a sustainable low cost breeding program for Nile tilapia that addresses both genetic and economic aspects of smallholder fish farmers in Kenya. First, Analytical Hierarchy Process Technique was used to define a breeding goal based on farmer’s preferences for traits. Farmers’ preferences for traits differed significantly depending on income and market orientation. Low and medium income farmers preferred harvest weight (HW) while high income farmers preferred growth (GR) and survival (S) traits. Grouping farmers according to market objective (fingerling production or fattening) showed that fingerling producers preferred GR and S while fattening farmers preferred HW and S. Consensus preference values were obtained using weighted goal programming and these values were used to derive desired gains for a breeding goal that takes into account farmers’ diverse backgrounds and preferences for traits. Secondly, the existence of genetic variation for traits of interest was investigated. Substantial additive genetic effects for HW, GR and shape traits were present that can be exploited through selection under low input production system. Heritability estimates for HW, GR and shape were 0.21 ± 0.03, 0.26 ± 0.04 and 0.12 ± 0.03 for mixed sex (nucleus) respectively. The calculation of economic values for breeding goal traits revealed that economic values for GR differed depending on the definition of the breeding goal and that selection for feed efficiency is the key factor to economic profitability of Nile tilapia breeding programs. A significant genotype by environment re-ranking was found for GR between the mixed sex nucleus and monosex production environments. Genotype by environment interaction (G x E) led to lower genetic gain for GR in production environment. Incorporating sib information from monosex production environment into the selection index resulted in a more accurate estimation of breeding values which increased genetic gain in growth. Using desired gain approach, weights for desired gains in harvest weight, growth rate and survival were derived that maximized genetic gains for these breeding goal traits. It is concluded that these results can be used to develop a sustainable centralized breeding program. However, a reliable well planned and organized decentralized strategy for dissemination of genetically improved fry of Nile tilapia to farmers is paramount.