Assessing environmental health risk is fraught with difficulties: often, the agents under study are weak pathogens, and exposures are poorly characterised. Direct studies in man are frequently inconclusive (or inconsistent) because of the limited sensitivity of epidemiological studies. Experimental studies raise questions about the validity of animal to man extrapolation. However, taking steps to protect public health is needed and impose the use of all available data. Remaining uncertainties lead the health risk assessor to make assumptions. Such a process implies as much transparency as possible, both for ethical reasons and to allow the public to be involved in the debate. A formalized approach has been proposed by various US agencies in charge of public health. One important matter is a strict separation between risk assessment and risk management. The procedure is divided into four portions: hazard assessment, exposure assessment, dose-response assessment which includes the study of risk at low doses--sometimes through the use of mathematical high to low dose extrapolation modelling--and risk characterisation.