Attributable risk is an important epidemiological index, and may be more relevant to many health planning situations than other indices such as the relative risk. This paper outlines the requirements of typical epidemiological data in order to estimate attributable risks, and illustrates the calculations for some common types of research study. The examples chosen have no special significance per se, but represent the main findings of several typical investigations reported in the epidemiological literature. The combination of estimates from several studies and extrapolation to other populations are discussed. Finally the concept of attributable risk as the causal effect of a risk factor is presented.