Introduction Samples for diagnostic tests for SARS-CoV-2 can be obtained from the upper (nasopharyngeal/oropharyngeal swabs) or lower respiratory tract (sputum or tracheal aspirate or broncho-alveolar lavage - BAL). Data from different testing sites indicates different rates of positivity. Reverse-transcriptase polymerase chain reaction (RT-PCR) allows for semi-quantitative estimates of viral load as time to crossing threshold (Ct) is inversely related to viral load. Objectives The objective of our study was to evaluate SARS-CoV2 RNA loads between paired nasopharyngeal (NP) and deep lung (endotracheal aspirate or BAL) samples from critically ill patients. Methods SARS-CoV-2 RT-PCR results were retrospectively reviewed for 51 critically ill patients from 5 intensive care units in 3 hospitals ; Addenbrookes Hospital Cambridge (3 units), Royal Papworth Cambridge (1 unit), and Royal Sunderland Hospital (1 unit). At the times when paired NP and deep lung samples were obtained, one patient had been on oxygen only, 6 patients on non-invasive ventilation, 18 patients on ECMO, and 26 patients mechanically ventilated. Results Results collected showed significant gradient between NP and deep lung viral loads. Median Ct value was 29 for NP samples and 24 for deep lung samples. Of 51 paired samples, 16 were negative (below limit of detection) on NP swabs but positive (above limit of detection) on deep lung sample, whilst 2 were negative on deep sample but positive on NP (both patients were on ECMO). Conclusions It has been suggested that whilst SARS-CoV1 tends to replicate in the lower respiratory tract, SARS-CoV2 replicates more vigorously in the upper respiratory tract. These data challenge that assumption. These data suggest that viral migration to, and proliferation in, the lower respiratory tract may be a key factor in the progression to critical illness and the development of severe acute respiratory syndrome (SARS). Factors which promote this migration should be examined for association with severe COVID-19. From a practical point of view, patients with suspected severe COVID-19 should have virological samples obtained from the lower respiratory tract where-ever possible, as upper respiratory samples have a significant negative rate.