The epidemic of bovine spongiform encephalopathy (BSE) has been the most expensive disaster ever to have befallen farming in the UK. It is believed to have led to a new form of spongiform encephalopathy in humans and as yet there is no way of knowing how many people will die of this disease. In order to curtail the BSE epidemic major decisions had to be made, often on the basis of inadequate scientific data. These data may have been derived from experiments using small sample numbers. Here we review some examples of where this has happened, sometimes with a beneficial outcome and sometimes with a misleading outcome. The identification of BSE as a new disease depended on precise neuropathological observation of a small number of cases rather than the obvious occurrence of large numbers of sick animals. Similarly, the recognition that BSE may have led to disease in humans was based on the neuropathological and clinical picture of new variant Creutzfeldt-Jakob disease (CJD) rather than on an increase in the number of cases of CJD in the UK. Early in the BSE epidemic the possibility that disease could be maternally transmitted from cow to calf was raised, mainly because of a belief that such transmission occurs in scrapie disease of sheep. But, we argue, the evidence for maternal transmission of scrapie, collected in the 1960s, was based on small numbers and is inadequate. Subsequent research has shown a very substantial genetic component in scrapie and epidemiological data show no excess risk in the offspring of affected ewes relative to the risk in the offspring of affected rams. An experiment to determine whether maternal transmission occurs in BSE was flawed and was unable to distinguish between maternal transmission and genetic susceptibility to environmental contamination. An assessment of the risk of BSE to humans depends on determining the levels of infectivity in tissues and transmissibility across species. Data on both of these are deficient, so it is not possible to predict how many people in the UK or elsewhere will become affected with new variant CJD in the next fifty years. The assessment of whether BSE could be transmitted to sheep and whether sheep therefore pose a risk to humans is hampered by a serious lack of evidence about the epidemiology of scrapie in the UK and elsewhere. The UK has paid a heavy price for the BSE epidemic but lessons should be learned from the experience. Every country should have a Specified Offals Ban even if it has no cases of BSE because, by the time it has, it will be too late. Furthermore, the occasional case of BSE should not be regarded as insignificant since it may be the harbinger of an epidemic in the making.