Drinkers in the general population who exceed the ‘safe limits’ advocated by various experts are undoubtedly at high risk of alcohol-related harm, yet they contribute only a minority to the total numbers of alcohol casualties, This relationship, the so-called ‘preventive paradox’ was explored in some detail using different criteria for ‘safe limits’ for total consumption, for frequency of drinking and for maximal daily consumption: two population surveys and a study of special groups noted for high intake were used. The results were in general mutually corroborative, even when different criteria for harm were employed. Certain methodological problems attendant on such analyses were explored. Finally it was shown that the gains from the universal adoption of the conventional ‘safe limits’ within a population would be matched by an across-the-board per capita reduction to about 70% of current intake. The implication was briefly reviewed.