A stubborn problem in alcohol epidemiology is that of standardization of unit of measurement. Consistent use of the 'standard drink' in research reports is hampered by difficulties in the assessment of the alcohol content of, particularly, self-reported drinks. Alcohol content of a drink depends on strength of the beverage and volume of the glass or container from which the beverage is taken. Both factors vary considerably between times, regions and individuals. Interview protocols and questionnaires rarely take into account the fact that people consume alcoholic beverages from a large variety of glasses and containers. In the present study the common presumption is tested of equality of alcohol content of standard and self-reported drinks. The test consisted of measuring the amount of wine, fortified wine and spirits people usually pour in the glass typical for the beverage type. The sample was drawn from the general Dutch population in 1985. The results show that on average self-reported drinks taken at home contained more than the presumed standard (10 g per drink). The deviation was highest for spirits (+26%), followed by fortified wines (+14%) and least for wine (+4%). There seemed to be a positive relationship between deviation from 'standard' and strength of the alcoholic beverage. This result is in line with data on the coverage of sales data: aggregate, survey-based spirits consumption shows the lowest coverage of sales. The effect of the difference between actual and presumed content of drinks on estimates of consumption is an overall increase of 7.5%, higher for women (+12%) than for men (+6%). Results are discussed with respect to the use of the concept of 'standard unit' in research protocols and health education campaigns.