Objectives. Amid the increasing number of global pandemic coronavirus disease 2019 (COVID-19) cases, there is a need for a quick and easy method to obtain a non-invasive sample for the detection of this novel coronavirus 2019 (SARS-CoV-2). We aimed to investigate the potential use of saliva samples as a non-invasive tool for the diagnosis of COVID-19. Methods. From 27 March to 4 April, 2020, we prospectively collected saliva samples and a standard nasopharyngeal and throat swab in persons seeking care at an acute respiratory infection clinic in a university hospital during the outbreak of COVID-19. Real-time polymerase chain reaction (RT-PCR) was performed, and the results of the two specimens were compared. Results. Two-hundred pairs of the samples were collected. Sixty-nine (34.5%) patients were male, and the median (interquartile) age was 36 (28-48) years. Using nasopharyngeal and throat swab RT-PCR as the reference standard, the prevalence of COVID-19 diagnosed by nasopharyngeal and throat swab RT-PCR was 9.5%. The sensitivity and specificity of the saliva sample RT-PCR were 84.2% [95% confidence interval (CI) 79.2%-89.3%], and 98.9% (95% CI 97.5-100.3%), respectively. An analysis of the agreement between the two specimens demonstrated 97.5% observed agreement (kappa coefficient 0.851, 95% CI 0.723-0.979; p <0.001). Conclusions. Saliva specimens can be used for the diagnosis of COVID-19. The collection method is non-invasive, and non-aerosol generating. Using a saliva sample as a specimen for the detection of SARS-CoV-2 could facilitate the diagnosis of the disease, which is one of the strategies that helps in controlling the epidemic.