BACKGROUND: Saliva has garnered great interest as an alternative specimen type for molecular detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Data are limited on the relative performance of different molecular methods using saliva specimens and the relative sensitivity of saliva to nasopharyngeal (NP) swabs. METHODS: To address the gap in knowledge, we enrolled symptomatic healthcare personnel (n = 250) from Barnes-Jewish Hospital/Washington University Medical Center and patients presenting to the Emergency Department with clinical symptoms compatible with coronavirus disease 2019 (COVID-19; n = 292). We collected paired saliva specimens and NP swabs. The Lyra SARS-CoV-2 assay (Quidel) was evaluated on paired saliva and NP samples. Subsequently we compared the Simplexa COVID-19 Direct Kit (Diasorin) and a modified SalivaDirect (Yale) assay on a subset of positive and negative saliva specimens. RESULTS: The positive percent agreement (PPA) between saliva and NP samples using the Lyra SARS-CoV-2 assay was 63.2%. Saliva samples had higher SARS-CoV-2 cycle threshold values compared to NP swabs (P < 0.0001). We found a 76.47% (26/34) PPA for Simplexa COVID-19 Direct Kit on saliva and a 67.6% (23/34) PPA for SalivaDirect compared to NP swab results. CONCLUSION: These data demonstrate molecular assays have variability in performance for detection of SARS-CoV-2 in saliva.