Risk analysis can usefully be applied to potential health and safety risks from technology and pollution. We examine historical examples of how knowledge about risk is developed, discussing the capabilities as well as the limitations of analyses to identify potential risks, establish causes, and estimate the probability that harmful effects will occur. One speaks of ''risk'' because, in any particular instance, an adverse outcome may or may not occur; causative factors skew the probabilities of different outcomes. Accordingly, the detection and measurement of risk must be derived from samples of instances. Real-world observations of risk may be supplemented with controlled experiments, and risks of interest must often be estimated indirectly through extrapolation from analogous circumstances. The ensuing uncertainty affects the clarity of ranking risks against one another and limits the usefulness of risk analysis alone as a means of resolving social conflicts about risk.
