ChIP-Seq technology, which combines chromatin immunoprecipitation (ChIP) with massively parallel sequencing, is rapidly replacing ChIP-on-chip for the genome-wide identification of transcription factor binding events. Identifying bound regions from the large number of sequence tags produced by ChIP-Seq is a challenging task. Here, we present GLITR (GLobal Identifier of Target Regions), which accurately identifies enriched regions in target data by calculating a fold-change based on random samples of control (input chromatin) data. GLITR uses a classification method to identify regions in ChIP data that have a peak height and fold-change which do not resemble regions in an input sample. We compare GLITR to several recent methods and show that GLITR has improved sensitivity for identifying bound regions closely matching the consensus sequence of a given transcription factor, and can detect bona fide transcription factor targets missed by other programs. We also use GLITR to address the issue of sequencing depth, and show that sequencing biological replicates identifies far more binding regions than re-sequencing the same sample.