Whether based on production, tax or sales statistics, estimates of alcohol consumption are subject to bias arising from record keeping procedures and from unrecorded consumption. Based on an analysis of the accuracy of consumption estimates derived from sales statistics in Ontario, two significant biases are identified. The first problem concerns the use of 16 per cent as the conversion factor for wine. As the consumption of 12 per cent table wines has increased, this figure is no longer appropriate and results in an over estimation of total alcohol consumption by about 2 per cent. The second major problem is the unrecorded home production of wine and wine brandy, which results in an underestimation of consumption by about 6 to 7 per cent. The available evidence concerning other types of unrecorded production indicates that they are not significant. It is concluded that sales data underestimate ‘true’ alcohol consumption in Ontario by about 4 to 6 per cent at this time.