One hundred and fourteen oat (Avena sativa L.) varieties of worldwide origin were evaluated for genetic diversity based on 77 molecular polymorphisms produced by eight selective AFLP primer combinations. Genetic similarity, calculated using the DICE coefficient, was used for cluster analysis and principal component analysis was applied. In addition population structure was explored to identify discrete subpopulations based on allele frequency. Although clustering and population structure showed relationships with region and country of origin, there was no obvious relationship to hull presence or hull colour. Oat varieties originating from European breeding programs showed less diversity than varieties originating from North and South America. Associations between AFLP markers and agronomic traits (grain yield, groat yield, panicle emergence, plant height, and lodging) as well as kernel quality traits (kernel weight, test weight, screening percent and groat percent) were also investigated. Marker-trait associations were tested using a naïve simple regression model and five additional models that account for population structure. Significant associations were found for 23 AFLP markers, with many of these affecting multiple traits. This study demonstrates that diversity can be significantly enhanced using a global collection, and provides evidence for marker-trait associations that can be validated in segregating populations and exploited through marker-assisted selection.