FLT3-internal tandem duplication occurs in 20-30% of acute myeloid leukemia and confers an adverse prognosis with its allelic ratio being a key risk stratifier. The US Food and Drug Administration recently approved FLT3 inhibitors midostaurin and gilteritinib in FLT3 mutation-positive acute myeloid leukemia. Historically, FLT3 was tested by fragment analysis, which has become the standard method endorsed by international guidelines. However, next generation sequencing is increasingly used at acute myeloid leukemia diagnosis given its ability to simultaneously evaluate multiple clinically informative markers. As FLT3-internal tandem duplication detection was known to be challenging by next generation sequencing and the results carry profound prognostic and therapeutic implications, it is important to thoroughly examine its performance in FLT3-internal tandem duplication detection and allelic ratio classification. In a comparative study with fragment analysis, we retrospectively reviewed our experience using a custom-designed, hybridization capture-based, targeted next generation sequencing panel. Among 7902 cases, FLT3-internal tandem duplication was detected in 335 with variable sizes (3-231 bp) and insertion sites. Fragment analysis was also performed in 402 cases, demonstrating 100% concordance in FLT3-internal tandem duplication detection. In 136 dual-tested, positive cases, 128/136 (94%) exhibited concordant high/low allelic ratio classifications. The remaining 6% showed borderline low allelic ratio by next generation sequencing. The two methods were concordant in FLT3-tyrosine kinase domain mutation detection at the hotspot D835/I836 targeted by fragment analysis. Furthermore, seven mutations which may benefit from FLT3 inhibitor therapy were detected by next generation sequencing, in regions not covered by fragment analysis. Our study demonstrates that using a hybridization capture-based chemistry and optimized bioinformatics pipeline, next generation sequencing can reliably detect FLT3-internal tandem duplication and classify its allelic ratio for acute myeloid leukemia risk stratification. Next generation sequencing also exhibits superior comprehensiveness in FLT3 mutation detection and may further improve personalized, targeted therapy in acute myeloid leukemia.