Rapid and simple point-of-care detection of SARS-CoV-2 is an urgent need to prevent pandemic. Reverse transcription loop-mediated isothermal amplification (RT-LAMP) can detect SARS-CoV-2 more rapidly than RT-PCR. Saliva is non-invasive specimen suitable for mass-screening, but data comparing utility of nasopharyngeal swab (NPS) and saliva in RT-LAMP test are lacking and it remains unclear whether SARS-CoV-2 could be detected by direct processing of samples without the need for prior RNA extraction saliva. In this study, we compared utility of saliva and NPS samples for the detection of SARS-CoV-2 by a novel RT-fluorescence LAMP (RT-fLAMP). The sensitivity and specificity of the RT-fLAMP with RNA extraction were 97% and 100%, respectively, with equivalent utility of NPS and saliva. However, sensitivity was decreased to 71% and 47% in NPS and saliva samples without RNA extraction, respectively, suggesting that RNA extraction process may be critical for the virus detection by RT-fLAMP.