Abstract Background Saliva samples are less invasive but not considered the gold standard for detecting SARS-CoV-2, and they are not validated for the Roche Cobas® Liat® platform. We aimed to evaluate the performance of a saliva sample compared to nasopharyngeal (NP) swab in the detection of SARS-CoV-2 on the Roche Cobas® Liat® platform at a research site in rural Guatemala. Methods Adults in an existing cohort study with influenza-like illness (ILI) provided clinical data and underwent NP swab (Copan) collection by trained nurses. Swabs were immediately placed into universal transport media (UTM), stored at 2-8 °C for less than 30 minutes and then tested on the Cobas Liat platform for SARS-CoV-2. Consenting subjects who had not eaten or drank in the last 2 hours were asked to provide a 5-mL saliva sample directly into an RNAse free container. The saliva sample remained at 2-8 °C for 24 hours, then diluted 1:2 with 0.85% saline (to reduce viscosity) and run on the same assay. We used descriptive statistics to compare the performance of saliva to NP swabs. Results Of the 28 subjects screened (1/10/22 - 4/26/22), 23 (82%) were consented and enrolled. The majority of subjects were male (78.3%) with a mean age of 31 years (range: 18-59 years). Of the 23 subjects enrolled, 14 (30.4%) reported fever and cough, 16 (34.8%) reported cough only, 14 (30.4%) reported fever only, and 2 (4.3%) reported cough and nasal congestion. The median symptom duration was 3.0 days (IQR: 1.5-5 days). Of the 23 subjects, 5 (23.7%) tested positive for SARS-CoV-2 in both sample types (NP swab and saliva); 3 (13%) were discordant, including 1 (4.3%) saliva-positive only and 2 (8.7%) NP-positive only. Compared to NP swab, preliminary performance of saliva in detection of SARS-CoV-2 included a sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of 71.4%, 93.7%, 87%, 83.3%, and 88.2%, respectively. The Fisher’s exact test p-value (0.003) shows concordance between both tests. Conclusion Our preliminary results show good precision between NP and saliva samples in detection of SARS-CoV-2 on the Roche Cobas® Liat® platform. Ongoing data collection will provide greater insight on the discordant results, but our findings support the continued use of saliva in the detection of SARS-CoV-2 in this population. Disclosures Molly Lamb, PhD, Pfizer: Grant/Research Support Edwin J. Asturias, MD, Curevac: DSMB Member|Fundacion para la Salud Integral de los Guatemaltecos: Board Member|Inovio: DSMB Member|Merck: Honoraria|Pfizer: Grant/Research Support Diva Calvimontes, n/a, Pfizer: Grant/Research Support Daniel Olson, MD, Fundacion para la Salud Integral de los Guatemaltecos: Board Member|Pfizer: Grant/Research Support|Roche: Grant/Research Support.