BackgroundAccurate measures of alcohol consumption are critical in assessing health harms caused by alcohol. In many countries, there are large discrepancies between survey-based measures of consumption and those based on alcohol sales. In England, surveys measuring typical alcohol consumption account for only around 60% of alcohol sold. Here, using a national survey, we measure both typical drinking and atypical/special occasion drinking (i.e., feasting and fasting) in order to develop more complete measures of alcohol consumption.MethodsA national random probability telephone survey was implemented (May 2013 to April 2014). Inclusion criteria were resident in England and aged 16 years or over. Respondents (n = 6,085) provided information on typical drinking (amounts per day, drinking frequency) and changes in consumption associated with routine atypical days (e.g., Friday nights) and special dinking periods (e.g., holidays) and events (e.g., weddings). Generalized linear modelling was used to identify additional alcohol consumption associated with atypical/special occasion drinking by age, sex, and typical drinking level.ResultsAccounting for atypical/special occasion drinking added more than 120 million UK units of alcohol/week (~12 million bottles of wine) to population alcohol consumption in England. The greatest impact was seen among 25- to 34-year-olds with the highest typical consumption, where atypical/special occasions added approximately 18 units/week (144 g) for both sexes. Those reporting the lowest typical consumption (≤1 unit/week) showed large relative increases in consumption (209.3%) with most drinking associated with special occasions. In some demographics, adjusting for special occasions resulted in overall reductions in annual consumption (e.g., females, 65 to 74 years in the highest typical drinking category).ConclusionsTypical drinking alone can be a poor proxy for actual alcohol consumption. Accounting for atypical/special occasion drinking fills 41.6% of the gap between surveyed consumption and national sales in England. These additional units are inevitably linked to increases in lifetime risk of alcohol-related disease and injury, particularly as special occasions often constitute heavy drinking episodes. Better population measures of celebratory, festival, and holiday drinking are required in national surveys in order to adequately measure both alcohol consumption and the health harms associated with special occasion drinking.Electronic supplementary materialThe online version of this article (doi:10.1186/s12916-015-0337-0) contains supplementary material, which is available to authorized users.