<p>Accurate species identification is a challenge in megadiverse regions, where morphology alone can lead to overestimation or underestimation of biodiversity. In this context, the incorporation of molecular approaches, such as DNA barcoding, has been useful in accurately estimating regional biodiversity, revealing cryptic diversity, and enabling identification by non-specialists. This study generated a molecular database of fishes for Parque Nacional dos Lençóis Maranhenses and adjacent areas, based on sequences of the mitochondrial cytochrome c oxidase subunit I (COI) gene. Haplotype analyses were conducted using Maximum likelihood (ML), and the delimitation of Operational Taxonomic Units (OTUs) combined the following methods: Kimura 2-parameter model (K2P), Assemble Species by Automatic Partitioning (ASAP) and Poisson Tree Processes (PTP) methods, using K2P ≥ 2% to resolve cases of incongruence between ASAP and PTP. A total of 59 species were recorded, distributed across 51 genera, 29 families, and 13 orders. In total, 258 sequences were analyzed (Two hundred and fifty-three generated in this study and five additional from GenBank), corresponding to the 59 species. K2P genetic distances revealed a clear separation between intraspecific and interspecific variation, with maximum and minimum values of 1.78% and 3.63%, respectively, resulting in a barcoding gap of 1.85% without overlap, demonstrating the high efficiency of the COI marker in species discrimination. Five new records for the area were identified, including two non-native species. The integration of molecular and morphological data enabled a revision of the ichthyofauna, including nomenclatural updates and the correction of identifications. These findings are essential for conservation strategies.</p>