SUMMARY Two aspects of the recent controversy about the health consequences of lead in petrol are considered. Firstly, the evidence is shown to be deficient, largely because the basic epidemiological principles of representative sampling, realistic measurement, and control of confounding variables were not followed so that valid conclusions cannot be drawn from most of the published studies. Secondly, the role of science appeared to be comprised by confusion between science per se and social policy. Relations between the two are explored, and it is concluded that confusing them reduces the contribution that science can make to effective social policy.The controversy about lead in petrol has been a curious phenomenon in a number of respects. It appeared to polarise scientific opinion to a degree that is unusual, and it suddenly subsided early in 1983. This last did not appear to be consequent on any new scientific findings or any new insight into available evidence, but it occurred within days of an announcement of a government commitment to reduce lead in petrol. We attempt to examine two aspects of this situation: firstly, the quality of the evidence behind the controversy, and secondly, the relation between science and policy making.