Saliva samples could be used as a non-invasive method to diagnose COVID-19. We aimed to assess the results of the reverse transcriptase-polymerase chain reaction (RT-PCR) of saliva specimens in the detection of COVID-19. We collected saliva and nasopharyngeal (NP) samples from consecutive COVID-19 suspects in Al-Fallujah Teaching Hospital, Anbar, Iraq from November 29, 2021 to February 15, 2022. The results of the two specimens were compared using RT-PCR. For the positive saliva tests, repetition of the test was undertaken at weekly intervals for four weeks from the time of the presentation. There were 55% men and 60% people ≤ 35 years. The majority of cases presented within 2-5 days (92%) and were of mild severity (89%). A hundred pairs of samples were taken. COVID-19 was diagnosed by NP swab RT-PCR in 56% and 31% of the saliva samples. The saliva samples had 100% sensitivity (95% confidence interval [CI] 60.4% e96.6%), 63.8% specificity (95% CI 96.1% e99.9%), and mild coefficient agreement (kappa coefficient = 0.522). The positive test for the saliva samples remained as such in all examined cases in the first and second weeks after the first test, 31/31 and 30/30, respectively. While half of them were positive in the third week (15/30). All cases became negative in the fourth week (0/15). We recommend not using the saliva swab as an alternative to the NP swab in the detection of the SARS-CoV-2 by RT-PCR. However, saliva sample can be used for the follow-up of the COVID-19 subjects, in children, elderly, and handicapped patients.