Quantitative information on alcohol consumption patterns derives from four possible sources: indirect measures, observational studies, aggregate consumption statistics, and sample surveys of general populations. The potentials and problems of each method are briefly discussed, with primary attention to the various traditions of survey questioning and data analysis. While medically oriented epidemiologists have often used only an overall drinking volume measure, social scientists have pointed to the importance also of variability in characterizing drinking, particularly in relation to social and casualty as opposed to chronic health problems with drinking. The dimensions of drinking patterns which might be relevant to hypothesized linkages of alcohol and cancer are discussed. It is suggested that measurement will need to extend beyond volume of drinking and may indeed involve studies of new kinds of dimensions in the patterning of drinking.