Scientists have made remarkable progress in dealing with technical challenges but not in dealing with society. Given that public concerns have grown, in the face of declining "real" risks, the common if simplistic tendency has been to blame public ignorance or irrationality and to argue that policy decisions should be based on quantitative risk estimates, effectively ignoring public concerns. Such assertions are superficially plausible, but they reflect fundamental misunderstandings of the nature of technological societies, as well as of the reasons behind declining scientific credibility and of actual strengths and weaknesses of risk assessment. Scientific credibility has been undermined not so much by shadowy enemies as by actions of self-proclaimed friends, and there are inherent limitations to the practical usefulness of risk assessment in policy disputes. If proposals for risk-based decision making were actually implemented, they could well lead not to increased credibility for specific technologies but to self-reinforcing losses of credibility for science and technology as a whole.