Copy number variants (CNVs) are thought to play an important role in the predisposition to autism spectrum disorder (ASD). However, their relatively low frequency and widespread genomic distribution complicates their accurate characterization and utilization for clinical genetics purposes. Here we present a comprehensive analysis of multi-study, genome-wide CNV data from AutDB (<ext-link ext-link-type="uri" xlink:href="http://mindspec.org/autdb.html" xlink:type="simple" xmlns:xlink="http://www.w3.org/1999/xlink">http://mindspec.org/autdb.html</ext-link>), a genetic database that accommodates detailed annotations of published scientific reports of CNVs identified in ASD individuals. Overall, we evaluated 4,926 CNVs in 2,373 ASD subjects from 48 scientific reports, encompassing ∼2.12×109 bp of genomic data. Remarkable variation was seen in CNV size, with duplications being significantly larger than deletions, (P = 3×10-105; Wilcoxon rank sum test). Examination of the CNV burden across the genome revealed 11 loci with a significant excess of CNVs among ASD subjects (P<7×10-7). Altogether, these loci covered 15,610 kb of the genome and contained 166 genes. Remarkable variation was seen both in locus size (20 - 4950 kb), and gene content, with seven multigenic (≥3 genes) and four monogenic loci. CNV data from control populations was used to further refine the boundaries of these ASD susceptibility loci. Interestingly, our analysis indicates that 15q11.2-13.3, a genomic region prone to chromosomal rearrangements of various sizes, contains three distinct ASD susceptibility CNV loci that vary in their genomic boundaries, CNV types, inheritance patterns, and overlap with CNVs from control populations. In summary, our analysis of AutDB CNV data provides valuable insights into the genomic characteristics of ASD susceptibility CNV loci and could therefore be utilized in various clinical settings and facilitate future genetic research of this disorder.