Since 1975, acceptance of pertussis vaccine has fallen from over 70% to 50% or less in most parts of Britain. This permits evaluation of a continuing natural experiment in which the frequency and severity of whooping cough can be compared those of adverse events following injections of pertussis vaccine. National data show an increase in notifications of whooping cough in most parts of Britain since 1975. Hospital admissions show considerable variation between areas with relatively high rates in some areas of deprivation but very low rates in more affluent areas even where vaccine-acceptance is around 50%. Deaths of infants with whooping cough have decreased steadily since 1900, the rate since 1975 being the lowest ever. Active epidemiological surveillance in Glasgow, with a population of 216,000 children and 13,000 births annually, shows that outbreaks and severe cases requiring admission to hospital were concentrated consistently in a few areas of deprivation. There is a significant correlation between vaccine-acceptance and hospital admission by district of residence but analysis of variance shows this effect to be less than that of overcrowding in households and other deprivation variables. In each of three outbreaks studied prospectively (1974-5, 78-78 and 82) about 30% of cases occurred in children who had received three doses of pertussis vaccine. Such vaccination had a significant protective effect in children aged 1-4 years but not in older children. There was no evidence of a herd immunity sufficient to protect infants below age for vaccination. Admissions to hospital decreased during the period 1970-83. There were no deaths attributable to proven or suspected infections with Bordetella pertussis during the period 1972-1983. No cases of encephalopathy, permanent brain damage or lung damage were detected in a follow up of all cases notified, surveyed or admitted to hospital between 1975 and 1982. Collectively, these national and local data provided estimates of the frequency of infection, complications of infection, admission to hospital and death in children with whooping cough for comparison with local, national and published estimates of the frequency and severity of adverse reactions, encephalopathy, permanent brain damage and death after injections of pertussis vaccine. It is concluded that, in children living in non-deprived circumstances in Britain, the risk of pertussis vaccine during the period 1970-83 exceeded those of whooping cough. In some deprived sectors, the risks from whooping cough might have been marginally higher but there was no evidence that this was associated with any increase in deaths or permanent disabilities.