Despite inter-individual variations in their diagnostic efficiency, dogs have been trained to investigate many human pathologies, especially cancer, diabetes, migraine, seizures and even infectious diseases. To this end, we performed a critical review and pooled analysis of current scientific literature on the performance of dogs trained for identifying severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-positive human specimens. We carried out an electronic search in PubMed, Scopus and Web of Science with the keywords "dog(s)" AND "sniffer" OR "scent" OR "smell" AND "SARS-CoV-2" OR "severe acute respiratory syndrome coronavirus 2" OR "coronavirus disease 2019" OR "COVID-19" within all fields, without date or language restrictions, to identify studies describing dogs' performance for identifying SARS-CoV-2 infected material. Three studies could be finally included in pooled analysis, totaling 17 dogs (47% females), aged between 0.5 and 12 years. The pooled diagnostic sensitivity was 0.88 (95% CI, 0.84-0.91; I2, 85.3%), the diagnostic specificity 0.99 (95% CI, 0.99-0.99; I2, 97.4%), whilst the area under the summary receiver operating characteristic curve (SROC) was 0.979 (standard error, 0.003). The notable performance observed in this pooled analysis would persuade us to suggest that adequately trained dogs could represent an intriguing and sustainable resource for purposes of rapid SARS-CoV-2 mass screening.