Foodborne hepatitis A infections have been considered as a major threat for public health worldwide. Increased incidences of hepatitis A virus (HAV) infection has been associated with growing global trade of food products. Rapid and sensitive detection of HAV in foods is very essential for investigating the outbreaks. Real-time RT-PCR has been most widely used for the detection of HAV by far. However, the technology relies on fluorescence determination of the amplicon and requires sophisticated, high-cost instruments and trained personnel, limiting its use in low resource settings. In this study, a robust, affordable, and simple assay, reverse transcription loop-mediated isothermal amplification (RT-LAMP) assay in combination with a bioluminescence-based determination of amplification in real-time (BART), was developed for the detection of HAV in different food matrices, including green onion, strawberry, mussel, and milk. The efficiencies of a one-step RT-LAMP-BART and a two-step RT-LAMP-BART were investigated for the detection of HAV in different food matrices and was compared with that of real-time RT-PCR. The sensitivity of the RT-LAMP-BART assay was significantly affected by Mg
2+
concentration (
P
< 0.05), in addition to primer quality. The optimal Mg
2+
concentration was 2 mM for one-step RT-LAMP-BART and 4 mM for two-step RT-LAMP-BART. Compared with cartridge-purified primers, HPLC-purified primers could greatly improve the sensitivity of the RT-LAMP-BART assay (
P
< 0.05). For detecting HAV in different food matrices, the performance of two-step RT-LAMP-BART was comparable with that of real-time RT-PCR and was better than that of one-step RT-LAMP-BART. The detection limit of the two-step RT-LAMP-BART for HAV in green onion, strawberry, mussel, and milk was 8.3 × 10
0
PFU/15 g, 8.3 × 10
1
PFU/50 g, 8.3 × 10
0
PFU/5 g, and 8.3 × 10
0
PFU/40 mL, respectively. The developed RT-LAMP-BART was an effective, simple, sensitive, and robust method for foodborne HAV detection.