High-throughput sequencing (HTS) technologies have brought tremendous improvements in the ability to detect plant viruses and have a great potential for application in virus routine diagnostics. The performance criteria of an HTS test need therefore to be estimated and compared to traditional virus indexing tests before it can be used in routine diagnostic. In this study, 78 Musa accessions previously indexed for viruses by molecular tests and/or electron microscopy were tested individually or in pools using an HTS protocol based on total RNA sequencing. The analytical sensitivity of HTS and RT-PCR was also compared by independent testing on serial dilutions of RNA extracts. In total, 136 libraries were sequenced in five batches, and the sequences analyzed for virus detection. The external alien control, a wheat sample infected by BYDV, monitored the contamination burden and determined an adaptative detection threshold. Overall, HTS test displayed a better analytical sensitivity compared to RT-PCR and a better inclusivity than the classical indexing protocol, as distant isolates and new viral species were only detected by the HTS test. Repeatability and reproducibility of virus detection were both of 100% although differences in number of sequencing reads per virus were observed between replicates. The diagnostic sensitivity was very high but false positive results were observed. Finally, the results also underlined the need of expert judgement in the interpretation of the results. In conclusion, the HTS test with an alien control and completed by expert evaluation has fulfilled the criteria of virus indexing protocol for Musa germplasm.