Human virus transmission through food consumption has been identified since many years and the international trade increases the risk of dissemination of viral pathogens. The development of metagenomic approach holds many promises for the surveillance of viruses in food and water. This work aimed to analyze norovirus diversity and to evaluate strain-dependent accumulation patterns in three oyster types by using a metagenomic approach. Different hexamer sets to prime cDNA were evaluated before capture-based approach to enhance virus reads recovery during deep sequencing. The study includes the use of technical replicates of artificially contaminated oysters and the analysis of multiple negatives controls. Results showed a clear impact of the hexamer set used for cDNA synthesis. A set of In-house designed (I-HD) hexamers, selected to lower mollusk amplification, gave promising results in terms of viral reads abundancy. However, the best correlation between CT values, thus concentrations, and number of reads was observed using random hexamers. Random hexamers also provided the highest numbers of reads and allowed the identification of sequence of different human enteric viruses. Regarding human norovirus, different genogroups and genotypes were identified among contigs longer than 500 bp. Two full genomes and six sequences longer than 3600 bases were obtained allowing a precise strain identification. The use of technical triplicates was found valuable to increase the chances to sequence viral strains present at low concentrations. Analyzing viral contamination in shellfish samples is quite challenging, however this work demonstrates that the recovery of full genome or long contigs, allowing clear identification of viral strains is possible.