Objectives:To investigate whether patterns of National Early Warning Scores (NEWS/NEWS2) in care homes during the COVID pandemic correspond with area-level COVID-19 death registrations from care homes. Study design: Longitudinal ecological study. Setting: 460 Care home units using the same software package to collect data on residents, from 46 local authority areas in England. Participants: 6,464 care home residents with at least one NEWS recording. Exposure measure: 29,656 anonymised person-level NEWS from 29/12/2019 to 20/05/2020 with component physiological measures: systolic blood pressure, respiratory rate, pulse rate, temperature, and oxygen saturation. Baseline values for each measure calculated using 80th and 20th centile scores before March 2020. Outcome measure: Time series comparison with Office for National Statistics (ONS) weekly reported registered deaths of care home residents where COVID-19 was the underlying cause of death, and all other deaths (excluding COVID-19) up to 10/05/2020. Results: Deaths due to COVID-19 were registered from 23/03/2020 in the study geographical areas. Between 23/03/2020 and 10/05/2020, there were 5,753 deaths (1,532 involving COVID-19 and 4,221 other causes). The proportion of above-baseline NEWS increased from 16/03/2020 and closely followed the rise and fall in COVID-19 deaths over the study period. The proportion of above-baseline oxygen saturation, respiratory rate and temperature measurements also increased approximately two weeks before peaks in care home deaths in corresponding geographical areas. Conclusions: NEWS may make a useful contribution to disease surveillance in care homes during the COVID-19 pandemic. Oxygen saturation, respiratory rate and temperature could be prioritised as they appear to signal rise in mortality almost as well as total NEWS. This study reinforces the need to collate data from care homes, to monitor and protect residents' health. Further work using individual level outcome data is needed to evaluate the role of NEWS in the early detection of resident illness.