BACKGROUND: The implementation of primary healthcare-based screening and advice that is effective in reducing heavy drinking can be enhanced with training. OBJECTIVES: Undertaking secondary analysis of the five-country ODHIN study, we test: the extent to which practice, provider and patient characteristics affect the likelihood of patients being screened and advised; the extent to which such characteristics moderate the impact of training in increasing screening and advice; and the extent to which training mitigates any differences due to such characteristics found at baseline. METHODS: A cluster randomized factorial trial involving 120 practices, 746 providers and 46 546 screened patients from Catalonia, England, the Netherlands, Poland, and Sweden. Practices were randomized to receive training or not to receive training. The primary outcome measures were the proportion of adult patients screened, and the proportion of screen-positive patients advised. RESULTS: Nurses tended to screen more patients than doctors (OR = 3.1; 95%CI: 1.9, 4.9). Screen-positive patients were more likely to be advised by doctors than by nurses (OR = 2.3; 95%CI: 1.4, 4.1), and more liable to be advised the higher their risk status (OR = 1.9; 95%CI: 1.3, 2.7). Training increased screening and advice giving, with its impact largely unrelated to practice, provider or patient characteristics. Training diminished the differences between doctors and nurses and between patients with low or high-risk status. CONCLUSIONS: Training primary healthcare providers diminishes the negative impacts that some practice, provider and patient characteristics have on the likelihood of patients being screened and advised. Trial registration ClinicalTrials.gov. Trial identifier: NCT01501552.