On 1 November, news organisations reported the global Covid-19 death toll had exceeded 5 million. But, as these articles highlight, this figure is likely to be a massive underestimate. Johns Hopkins University collates official daily statistics on Covid deaths, but there is no unified global definition: Belgium’s high reported death rate partly reflects its including all probable Covid deaths in all settings, while Hungary only publishes hospital deaths with a positive test. Turkmenistan and North Korea have, apparently, not experienced a single Covid death. The UK surveillance death count based on positive tests was around 140,600 on 1 November, but the number of death certificates mentioning Covid is considerably higher, at about 164,500. Even that higher figure could be too low, with under-diagnosis of this new disease in early months. A more robust method is to estimate the “excess” over the number who would have died in normal times. Of course, pandemics may increase deaths from other causes, say through disruption to healthcare, although restrictions and behavioural changes can reduce deaths too. The Office for National Statistics uses an average between 2015 and 2019 as a baseline and in recent weeks finds a worrying excess of many hundreds of deaths not involving Covid. However, the Institute and Faculty of Actuaries adjusts for a changing population to create an alternative baseline, which yields fewer than 100 non-Covid “extra” deaths in the latest week. To get an uncertain answer to the right question, the Economist built a model to estimate global excess deaths. Russia has reported 230,000 Covid deaths, but may have about four times as many excess deaths. Devastated by the Delta variant, India’s estimated excess mortality is around 10 times higher than its Covid surveillance death count of 460,000, although the uncertainty interval is wide (three to 16 times higher). Inadequate death registration hampers this analysis – the UN statistics division estimates fewer than seven in 10 territories have over 90% coverage. The Economist’s model estimates 10 to 19m extra deaths around the world during the pandemic. Five million deaths is a grim milestone, but humanity passed that long ago. David Spiegelhalter is chair of the Winton Centre for Risk and Evidence Communication at Cambridge. Anthony Masters is statistical ambassador for the Royal Statistical Society