White lupin (Lupinus albus L.) is one of four economically important species of the Lupinusrngenus, and has been traditionally cultivated for several thousand years along the Nile valley,rnincluding Ethiopia. Lupins are known to perform multifaceted functions, such as for human foodrnand beverage, livestock feed, ecological importance, pharmaceutical values and socialrncontributions. In Ethiopia, white lupin has been sustaining quite long in the farming system, andrnis produced exclusively by smallholder subsistence farmers. However, despite it has long beenrnproduced in the country, the crop received little attention by different development actors andrnhave several undesirable characteristics. To address these, setting up practical lupin breedingrnprogram targeting the aforementioned major lupin production constraints is quite essential andrnnecessary. Hence, the present study attempted to avail various pertinent socioeconomic andrngenetic and/or breeding information which are fundamental to realize lupin improvement inrnthe country.rnDetailed baseline survey involving 303 households sampled from white lupin production areasrnof north western Ethiopia was conducted, to ascertain the extent of lupin productionrnconstraints and document farmers experiences and practices on lupin cropping, processing andrnutilization and marketing. Study results identified the major production practices and constraints to lupin production in the areas. The results would form a useful guide for therndevelopment of wellâ€tailored breeding objectives for the improvement of white lupin forrnEthiopian farmers and consumers. This approach is useful not only to document farmersrnexperiences and practices, but also ensures participation of farmers to develop demand ledrnlupin technologies.rnA phenotyping experiment aiming at characterizing the landraces using agronomic andrnphenological traits which comprises 143 landrace accessions was under taken at Merawi,rnEthiopia. Further characterization of genetic diversity and population structure of 212 landracesrnusing 15 polymorphic SSR markers were done. Another experiment with objectives to evaluaternthe performance and stability of white lupin landraces in different locations; and characterizernwhite lupin growing environments in Ethiopia was conducted at six different locations in northrnwestern Ethiopia. The phenotyping and overâ€locations experiments were considered tornunderstand the relationships among traits, and to document trait profile of white lupin landraces.rnPhenotypic characterization revealed that Ethiopian white lupin landraces were significantlyrndifferent for most of the traits studied, and a significant number of local accessions performedrnas high as 5 metric tonnes per hectare of grain yield. Cluster analysis showed that landracesrnwere grouped into 17 clusters of different sizes, of which five were singletons. Some landracesrnwere grouped together regardless of their geographic origin. On the other hand, landraces fromrnAwi, South Gondar and West Gojam were distributed over many clusters. Genetic distancesrnbetween many pairs of clusters were significant, justifying crosses between parents from themrnwould be desirable genetic recombinations. Molecular characterization further revealed the genetic diversity vested on the landraces. ThernSSR markers revealed 98 from 212 landraces, with an average of 6.5 alleles per locus. Thernaverage gene diversity was 0.31. Twenty eight landraces harbored one or more private allelesrnfrom the total of 28 private alleles identified in the 212 white lupin accessions. Seventyâ€sevenrnrare alleles with a frequency of less than 5% were identified and accounted for 78.6% of therntotal allele detected. Analysis of molecular variance (AMOVA) showed that 92% of allelicrndiversity was attributed to individual accessions within populations while only 8% wasrndistributed among populations. At 70% similarity level, the UPGMA dendrogram resulted in thernformation of 13 clusters comprised of 2 to 136 landraces, with the two control genotypes andrnfive landraces remaining distinct and ungrouped. Population differentiation and geneticrndistance were relatively high between Gondar and populations collected by Australians. Highrnlevel of gene flow (Nm), ranging from 10.60 to 31.46, was detected between the four majorrnpopulations namely West Gojam, Awi, East Gojam and Gondar. A modelâ€based populationrnstructure analysis divided the white lupin landraces into two populations. All Ethiopian whiternlupin landrace populations, except most of the landraces collected by Australians and somernfrom Awi, were grouped together with significant admixtures. The study also suggested that 34,rnas core collections, were sufficient to retain 100% of SSR diversity.rnHigher heritability and genetic advance as percent of mean was observed for grain yield,rnindicating the possibility of improving this trait through selection. Different patterns ofrnassociations and accession by trait interactions were observed in different environments.rnHowever, genotype by trait biplots consistently indicated that grain yield had positive associations with most of the traits; especially, with number of pods per plant, plant height andrnseeds per pod. The study identified some accessions with desirable performances as good forrnspecific trait and/or trait groups that could be considered as sources of genes for the traits theyrnhave best performed.rnThe genotype by environment interaction study depicted that the white lupin landraces studiedrnhad differential performances at different test locations implying the presence of crossoverrninteraction. The first two principal components (PC1=41.6% and PC2=21.8%) of the genotypernplus genotype by environment interaction (GGE) explained 63.4% of the GGE sum of squares.rnTwo white lupin growing megaâ€environments were identified in north western Ethiopia. All testrnlocations were found to be representative with different degrees of reliability whereby FenoternSelam and Dibate were found to be most representative. In addition, all test locations, exceptrnMandura and Injibara had generally similar and good discriminating power. Fenote Selam andrnDibate were found to be the most representative and discriminating environments and arerncharacterized as most desirable test locations for white lupin improvement in north westernrnEthiopia. Genotype 2 (G2) was found to be the highest yielding and most stable landrace acrossrnthe test environments, and hence identified as most desirable genotype for production.rnKey words: Discriminating power, Ethiopian farming system, Farmers' experiences, Genotypernby location interaction, Genotype by Trait, Landrace populations, Megaâ€environment,rnRepresentativeness, White lupin.