Just as an imperfect ruler can measure accurately if its error is constant and known, serology need not have perfect sensitivity and specificity to estimate COVID's prevalence, write Drs Jha and Murthy.