AIMS: To identify communication characteristics of patients and counsellors during brief alcohol intervention (BAI) which predict changes in alcohol consumption 12 months later. METHODS: Tape-recordings of 97 BAI sessions with hazardous drinkers were analysed using the Motivational Interviewing Skill Code (MISC). Outcome measures were (i) baseline to a 12-month difference in the weekly drinking quantity, and (ii) baseline to a 12-month difference in heavy drinking episodes per month. Bivariate analyses were conducted for all MISC measures, and significant variables were included in multiple linear regression models. RESULTS: Patient communication characteristics (ability to change) during BAI significantly predicted the weekly drinking quantity in the multiple linear regression model. There were significant differences for some of the counsellor skills in bivariate analyses but not in the multiple regression model adjusting for patients' talk characteristics. Changes in heavy drinking showed no significant association with patient or counsellor skills in the multiple linear regression model. CONCLUSION: Findings indicate that the more the patient expresses ability to change during the intervention, the more weekly alcohol use decreases. The role of the counsellor during the interaction, and influence on the outcomes was not clearly established. Implications for BAI and related research are discussed.