Compared to RT-PCR, lower performance of antigen detection assays, including the Lumipulse G SARS-CoV-2 Ag assay, may depend on specific testing scenarios. We tested 594 nasopharyngeal swab samples from individuals with COVID-19 (RT-PCR cycle threshold [Ct] values ≤40) or non-COVID-19 (Ct values ≤40) diagnoses. RT-PCR positive samples were assigned to diagnostic, screening, or monitoring groups of testing. With a limit of detection of 1.2 × 104 SARS-CoV-2 RNA copies/ml, Lumipulse showed positive percent agreement (PPA) of 79.9% (155/194) and negative percent agreement of 99.3% (397/400), whereas PPAs were 100% for samples with Ct values of <18 or 18–<25 and 92.5% for samples with Ct values of 25–<30. By three groups, Lumipulse showed PPA of 87.0% (60/69), 81.1% (43/53), or 72.2% (52/72), respectively, whereas PPA was 100% for samples with Ct values of <18 or 18–<25, and was 94.4%, 80.0%, or 100% for samples with Ct values of 25–<30, respectively. RT-PCR positive samples were also tested for SARS-CoV-2 subgenomic RNA and, by three groups, testing showed that PPA was 63.8% (44/69), 62.3% (33/53), or 33.3% (24/72), respectively. PPAs dropped to 55.6%, 20.0%, or 41.7% for samples with Ct values of 25–<30, respectively. All 101 samples with a subgenomic RNA positive result had a Lumipulse assay’s antigen positive result, whereas only 54 (58.1%) of remaining 93 samples had a Lumipulse assay’s antigen positive result. In conclusion, Lumipulse assay was highly sensitive in samples with low RT-PCR Ct values, implying repeated testing to reduce consequences of false-negative results.