Exposure to cadmium can give rise to a number of adverse health effects. In human risk assessment it is important to identify effects that occur at relatively low exposures, i.e., the critical effects, which are crucial for preventive action. The effect of cadmium on the kidney, namely, tubular dysfunction, has been regarded as the critical effect in long-term human exposures. Quantitative dose-response data can be derived from a risk model based on metabolic parameters and observed relationships between cadmium concentrations in the renal cortex and the occurrence of proteinuria in industrial workers. There is reasonable agreement between this model and epidemiological observations. Low-dose-low-incidence extrapolations have been made based on this model. However, experimental and epidemiological studies are providing increasing evidence that cadmium is carcinogenic and this serious effect, considered to be stochastic in character, may reasonably be considered as a critical effect. A quantitative evaluation is difficult with the available data. The preventive action usually taken for substances which are considered to be carcinogenic is to limit the use of and exposure to cadmium as much as possible, thus achieving exposure levels for cadmium lower than those giving rise to kidney damage.