OBJECTIVE The goal of this study was to examine how well per capita alcohol consumption figures derived from beverage sales data relate to changes over time in survey-based measures of drinking patterns. It was expected that strong associations would be found among these various measures of consumption. METHODS Data from 12 household surveys conducted in Ontario between 1977 and 1997 provided information on: percentages of drinkers; daily drinkers; those drinking five or more drinks at a sitting weekly; those reporting two or more alcohol-related harms; and average number of drinks per week. These variables were then correlated with per capita consumption. RESULTS Significant correlations were found only between per capita consumption and percentage of daily drinkers, and between percentage of drinkers and average number of drinks per week. CONCLUSIONS The relationship of per capita consumption to survey measures of drinking is weak. The absence of consistent associations over time between per capita consumption and survey measures may be attributable to the small number of available data points or to increases in unrecorded consumption. Further research is needed to verify and explain these results.