Aims. To examine the dose-response relationship between self-reported alcohol consumption and levels of self-reported negative consequences of drinking. Design. Data from 10 general population random sample surveys over the years 1990-1997 were combined and responses were plotted and subjected to regression analysis. Setting. Auckland, a city of approximately 1 million people in the North Island of New Zealand. Participants were interviewed in their homes using a computer-assisted telephone interviewing system. Participants. General population sample of 11 817 aged 14-65 years, representative of the Auckland population. Measurements. Frequency of experience of 14 negative consequences; annual volume of alcohol consumed; frequency of drinking larger quantities of alcohol. Results. Three different patterns of relationship between consequences and consumption were found for different consequences. These differed between the prevalence and the frequency of consequences but were similar for two different measures of consumption, annual volume and larger quantity drinking. Analysis of the frequency of experienced consequences found that the risk curves for the most common consequences approximated a straight line and the effects at low volume intake were due to those drinking larger quantities. Three less common consequences clearly showed a concave curve suggesting a threshold effect, with effects beginning at about 20 litres per annum of absolute alcohol. Conclusion. The different relationships between consumption and consequences imply that some consequences occur only once a very heavy volume of drinking is reached, while others show a direct relationship with consumption, reflecting that volume of alcohol consumed is closely related to the quantities consumed.