Severe acute respiratory coronavirus 2 (SARS-CoV-2) testing reagents are expected to become scarce worldwide. However, little is known regarding whether pooling of samples accurately detects SARS-CoV-2. To validate the feasibility of pooling samples, serial dilution analysis and spike-in experiments were conducted using synthetic DNA and nucleic acids extracted from SARS-CoV-2-positive and -negative patients. Furthermore, we studied 1000 individuals, 667 of whom were “healthy” individuals (195 healthcare workers and 472 hospitalized patients with disorders other than COVID-19 infection), and 333 infection-suspected patients with cough and fever. Serial dilution analysis showed a limit of detection of around 10–100 viral genome copies according to the protocol of the National Institute of Infectious Diseases, Japan. Spike-in experiments demonstrated that RT-qPCR detected positive signals in pooled samples with SARS-CoV-2-negative and -positive patients at 5-, 10-, 20-fold dilutions. By screening with this pooling strategy, by the end of April 2020 there were 12 SARS-CoV-2-positive patients in 333 infection-suspected patients (3.6%) and zero in 667 “healthy” controls. We obtained these results with a total of 538 runs using the pooling strategy, compared with 1000 standard runs. In a prospective study, we successfully detected SARS-CoV-2 using 10- to 20-fold diluted samples of nasopharyngeal swabs from eighteen COVID-19 patients with wide ranges of viral load. Pooling sample is feasible for conserving test reagents and detecting SARS-CoV-2 in clinical settings. This strategy will help us to research the prevalence infected individuals and provide infected-status information to prevent the spread of the virus and nosocomial transmission.