OBJECTIVE: Although the effect of alcohol assessment in medical settings has received attention, the longitudinal study of such efforts has been restricted to studying a single assessment/intervention dose. Such interventions can be recurrent and have effects on subsequent problem drinking. METHOD: A sample of problem drinkers in the general population (n = 672) and with admissions to chemical-dependency programs (n = 926) was interviewed at baseline and 1, 3, 5, and 7 years later. At each wave, respondents were asked about their drinking, their medical visits, and the intensity of the medical contact (whether during the visit they were asked about their drinking and, if so, whether they received or were referred to alcohol treatment). RESULTS: Rates of problem drinking declined over time, from 48% at the 1-year follow up to 38% at the 7-year follow-up. Problem drinkers were more likely at each wave to receive or be referred to treatment. Alcohol and drug severity increased with more intensive medical-contact types over time. Predicting subsequent problem drinking status from prior intensity of medical contact, odds of problem drinking at subsequent waves decreased with time, age, and prior drug severity while increasing with volume and alcohol severity. Odds of problem drinking were lower among prior problem drinkers receiving assessment and treatment/referral, compared with the assessed-only group. Examined separately, this effect was found only for those drinkers with lower volumes (average < 0.5 drinks/day). Conclusions: Alcohol assessment may be effective in reducing problem drinking but may be most effective among the non-heaviest drinkers.