INTRODUCTIONHigh-density single nucleotide polymorphism (SNP) genotyping arrays recently have been used for copy number variation (CNV) detection and analysis, because the arrays can serve a dual role for SNP- and CNV-based association studies. They also can provide considerably higher precision and resolution than traditional techniques. Here we describe PennCNV, a computational protocol designed for CNV detection from high-density SNP genotyping data. This protocol extracts allele-specific signal intensities from genotyping arrays, and then integrates information on SNP spacing and SNP allele frequencies to generate CNV calls by a hidden Markov model (HMM) algorithm. Analyses of CNVs from SNP genotyping arrays will provide a more comprehensive view of genome variation, and complement current genome-wide association studies in identifying disease susceptibility loci.