One of the primary tools for diagnosing COVID-19 is the nucleic acid-based real-time reverse transcriptase-polymerase chain reaction (RT-PCR) test performed on respiratory specimens. The detection rate of SARS-CoV-2 in lower respiratory specimens (such as sputum) is higher than that for upper respiratory specimens (such as nasal and pharyngeal swabs). However, sputum specimens are usually quite viscous, requiring a homogenization process prior to nucleic acid (NA) extraction for RT-PCR. Sputum specimens from COVID-19 and non-COVID-19 patients were treated with four commonly used reagents—saline, N-acetyl-L-cysteine (NALC), proteinase K (PK), and dithiothreitol (DTT), prior to NA extraction. These reagents were then compared for their performance in diagnosing COVID-19 in real clinical practice. The detection rate of SARS-CoV-2 in PK- or DTT-treated sputum was comparable, and higher than that in sputum treated with NALC or saline. While there was a 4.8% (1/21) false negative rate for the PK- and DTT-treated sputum, neither treatment showed any false positive cases among patients with non-COVID diseases. Moreover, sputum pretreated with saline, NALC, PK or DTT showed higher detection rates of SARS-CoV-2 as compared to pharyngeal swabs. Taken together, we provide direct evidence recommending the use of PK or DTT to pretreat sputum samples to facilitate SARS-CoV-2 detection by clinical laboratories. Moreover, our methods should help to standardize the procedure of processing sputum specimens and improve the ability to detect SARS-CoV-2 in these samples.