A hypothesis on the effects of level and duration of alcohol misuse on individual risks for liver cirrhosis is proposed. The hypothesized risk function is a generalized version of the exponential law of other researchers. Some predictions are derived and tested against two types of aggregate data--the age structure of cirrhosis mortality and time-series data for the general population. It is shown that both types of data are compatible with the proposed risk function. It is also shown that the fact that extensive alcohol misuse will result in death from cirrhosis only after a considerable number of years does not contradict the observation that changes in per capita consumption normally have an instantaneous effect on cirrhosis mortality.