One big limitation of computational tools for analyzing ChIP-seq data is that most of them ignore non-unique tags (NUTs) that match the human genome even though NUTs comprise up to 60% of all raw tags in ChIP-seq data. Effectively utilizing these NUTs would increase the sequencing depth and allow a more accurate detection of enriched binding sites, which in turn could lead to more precise and significant biological interpretations. In this study, we have developed a computational tool, LOcating Non-Unique matched Tags (LONUT), to improve the detection of enriched regions from ChIP-seq data. Our LONUT algorithm applies a linear and polynomial regression model to establish an empirical score (ES) formula by considering two influential factors, the distance of NUTs to peaks identified using uniquely matched tags (UMTs) and the enrichment score for those peaks resulting in each NUT being assigned to a unique location on the reference genome. The newly located tags from the set of NUTs are combined with the original UMTs to produce a final set of combined matched tags (CMTs). LONUT was tested on many different datasets representing three different characteristics of biological data types. The detected sites were validated using de novo motif discovery and ChIP-PCR. We demonstrate the specificity and accuracy of LONUT and show that our program not only improves the detection of binding sites for ChIP-seq, but also identifies additional binding sites.