Introdução: Laboratory diagnosis of viral diseases is limited by the absence of isolation systems and the availability of validated methods, which allow the identification of some known pathogens and are non-existent for most emerging viruses. Objetivo(s): To standardize and validate a metagenomic NGS methodology for the diagnosis of viral infections. Material e Métodos: Training dataset was used for a benchmark to test different bioinformatics workflows and parameters to optimize time and number of pass reads during basecalling and demultiplex steps (Guppy Basecaller-Barcoder, ONT). Samples of viral isolates and clinical samples of arboviruses, acute respiratory tract infections, and chronic infections caused by viral hepatitis and HIV were analyzed. Viral RNA was extracted using column method (QIAGEN). The eluate was digested with DNAses to remove DNA genomes. The cDNA was synthesized and amplified by sequence- independent, singleprimer amplification (SISPA) for sequencing in a MinION device (ONT). Raw data was filtered from scratch using qscore9, length 100bp and DUST algorithm to finding low-complexity regions in DNA sequences. Minimap/Samtools was used for align GRCh38 and remove, and Racon for read correction. The taxonomic classification was performed using Kraken2. A cut-off of relative abundance of the number of reads >0.5% was used for virome analysis. Libraries failed due to one of these errors: initial lab methods standardization, inability to demultiplex and absence of virome results in all samples on the library. Resultados e Conclusão: Basecalling was optimized for both fast and hac models, but fast results were preferred for minimizing expended time (1.36 hs vs 30.71 hs) and keeping 84% of pass reads. Stringent demultiplex filter was used recovering 25% of the pass reads. Preliminary results showed eighteen(72%) out of 25 samples were successfully diagnosed: CHIKV(3), CHIK+MAYV(1), DENV(4), ZIKV(2), HCV(1), HIV(2), SARS-Cov-2(4), and GBV-C(1). Five were undetectable and two were undetermined. A quicker and less stringent workflow using a local database and taxonomic classification with BLAST was also evaluated with equivalent efficiency. There are complementary experiments in progress and the data is still being analyzed. MinION was able to perform virome analysis and diagnosis of viral infections with good accuracy. Advances in this methodology may allow its use in routine diagnosis, surveillance and discovery of emerging agents.