Several commercial nucleic acid amplification tests (NAAT) for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) have been developed. We used six kits available in Japan in 13 NAAT-positive specimens with crossing point values >36 and 7 NAAT-negative specimens from patients with coronavirus disease (COVID-19); their results were compared. Specimens positive in ≥2 assays were considered true positive and examined for concordance with specimen results. The SARS-CoV-2 Detection Kit -Multi- (Toyobo M) (Toyobo, Osaka, Japan) using extracted RNA had the highest concordance (κ 1.00). This was followed by the cobas® SARS-CoV-2 (Cobas) (Roche, Basel, Switzerland) (κ 0.79). There was a weak correlation between number of negative results for each kit and days between onset and testing (Spearman rank correlation: ρ= 0.44; p<0.05). We believe that the variations in results among kits for specimens with low viral loads should not be problematic when these kits are used for screening infectious patients because these variations are more likely to be observed in specimens tested many days after onset (i.e., those that have lost their infectivity). However, for suspected late-stage COVID-19 with a low viral load, it may be better to use a test such as Toyobo M or Cobas.