Never as with the present pandemics, numbers and the attendant activities of measuring and modelling have taken centre-stage. Yet these numbers, often delivered by academicians and media alike with extraordinary precision, rely on a rich repertoire of assumptions, including forms of bias, that can significantly skew both the numbers per se and the trust we repose in them. We discuss the issue in relation to a particular case relative to the numbers on excess mortality during the first wave of the Covid-19 pandemic in Italy. We conclude with some considerations about the use of science at the science policy interface in situations where facts are uncertain, stakes high, values in dispute and decision urgent.