Purpose: Compared to nasopharyngeal/oropharyngeal swabs (N/OPS-VTM), non-invasive saliva samples have enormous potential for scalability and routine population screening of SARS-CoV-2. In this study, we investigate the efficacy of saliva samples relative to N/OPS-VTM for use as a direct source for RT-PCR based SARS-CoV-2 detection. Methods: We collected paired nasopharyngeal/oropharyngeal swabs and saliva samples from suspected positive SARS-CoV-2 patients and tested using RT-PCR. We used generalized linear models to investigate factors that explain result agreement. Further, we used simulations to evaluate the effectiveness of saliva-based screening in restricting the spread of infection in a large campus such as an educational institution. Results: We observed a 75.4% agreement between saliva and N/OPS-VTM, that increased drastically to 83% in samples stored for less than three days. Such samples processed within two days of collection showed 74.5% test sensitivity. Our simulations suggest that a test with 75% sensitivity, but high daily capacity can be very effective in limiting the size of infection clusters in a workspace. Guided by these results, we successfully implemented a saliva-based screening in the Bangalore Life Sciences Cluster (BLiSC) campus. Conclusion: These results suggest that saliva may be a viable alternate source for SARS-CoV-2 surveillance if samples are processed immediately. Although saliva shows slightly lower sensitivity levels when compared to N/ OPS-VTM, saliva collection is logistically advantageous. We strongly recommend the implementation of salivabased screening strategies for large workplaces and in schools, as well as for population-level screening and routine surveillance as we learn to live with the SARS-CoV-2 virus.