The unique evolutionary history of domesticated dogs ( Canis lupus familiaris), including strong artificial selection, population bottlenecks, and inbreeding, has resulted in over 400 genetically distinct breeds that make them well suited for addressing fundamental questions in population genetics, evolution, and the genetic architecture of phenotypic variation. It is important to ascertain, to the fullest extent, the range and characteristics of all genetic variants to be able to make accurate analyses and correctly measure the effects of genetic changes on the wide assortment of canine phenotypes. Structural variation is as an important and abundant source of genetic and phenotypic variation. Here I describe the first systematic and genome-wide analysis of structural variation in the modern domesticated dog, Canis lupus familiaris, which exhibits considerable morphological, physiological, and behavioral variation. Through computational analyses I identify segmental duplications in the publicly available canine reference sequence. Experimentally I identify copy number variants (CNVs) associated with these segmental duplications in a panel of diverse dogs. Additionally, using a subset of this panel I further identify genome-wide CNVs. In total I find thousands of individual CNVs that cover approximately 900 distinct regions of the canine genome. I genotype over 60 individual dogs from 12 different breeds for copy number changes in these distinct regions and determine breed distributions of CNVs. I find that appreciable differences exist in between the breeds in terms of CNV content, but also there exists substantial CNV differences within the breeds themselves, highlighting the unique and active nature of copy number changes. In total I find over 600 diverse genes or predicted genes to be within or near CNV regions that could have functional or phenotypic impacts. My results provide insights into mechanisms of canine genome evolution and generate a valuable resource for future evolutionary and phenotypic studies.