Aim. To study the effect of changes in per capita alcohol sales and indicators of alcoholism treatment on admissions to inpatient care and mortality for liver cirrhosis and alcoholism, alcohol intoxication and alcohol psychosis (AAA). Design. Bivariate and multivariate time series analyses was conducted by applying the ARIMA-modelling technique. Setting and participants. All analyses were conducted on quarterly data from Stockholm County 1980-94 with a population of 1.7 million people. Measurements. Data on sales of alcohol and disulfiram/calcium carbimide were used as input variables. Inpatient data (from the Stockholm Inpatient Care Register) and mortality dam (from the Cause of Death Register) on all cases with alcoholism, alcohol psychasis and alcohol intoxication (AAA) and liver cirrhosis as underlying or contributory diagnoses were used as output variables. Findings. Alcohol sales affected the cirrhosis rate. For cirrhosis mortality, but not for cirrhosis admissions, the effect was not only direct but also distributed over time. Significant direct and time lag effects of alcohol sales on both AAA series and cirrhosis admissions were found only during earlier, shorter periods, e.g. 1980-90. All four output series showed significant effects of sales of disulfiram/calcium carbimide and were the only significant predictors for the two AAA endpoints for the whole study period. Conclusions. These results suggest that to reduce the rate of alcohol-related problems caused by socially deteriorated and severely alcohol-dependent subjects (i. e. AAA), reduction of overall consumption should be complemented by treatment of alcohol-dependent subjects.