For chemical pollutants, health risk assessment of long-term exposure is usually best realized through an epidemiologic approach which attempts to link cumulative levels of exposure to the potential for occurrence of early adverse effects. For some chemicals, however, the frequency of peak exposures may be more relevant for assessing the health risk than the integrated dose. In very few circumstances, biological exposure indices directly reflect the cumulative dose (e.g. PCB in blood). More frequently they are indicators of short-term interval dose but provided they have been measured with a sufficient frequency, their integration over the duration of exposure may represent a valid surrogate of the cumulative dose. This has been clearly demonstrated for lead or cadmium in blood. The selection of the appropriate biological effect markers for the study of the dose-effect/dose-response relationships is frequently a controversial issue when information on the mechanism of action of the pollutant is insufficient. In this case, the study of the health significance of the observed biological changes may be required for assessing a meaningful no-adverse-effect level. For example, in adult male workers moderate exposure to lead may affect the synthesis of vasodilatory prostaglandins in the kidney but presently there is no indication that this effect should be taken into account to define the acceptable occupational exposure level to lead because it is not associated with an impairment of the hemodynamic response of the kidney to an acute protein load. On the contrary, a low-molecular-weight proteinuria induced by cadmium may be predictive of an increased age-related decline of the glomerular filtration rate. Although the use of early biological effect markers for the study of the dose-effect or dose-response relationships in humans is probably less affected by selection biases than morbidity data, the possibility of such an interference cannot be excluded. For example, in the general population, the tubulotoxic effects of cadmium may occur at a lower body burden of the metal than in adult male workers. Whatever the adverse biological effect considered, the application of an uncertainty factor remains justified when extrapolating a no-effect level from adult male workers to the general population.