OBJECTIVES: This research was designed to find out the key attribute for benefit-risk assessment using swing weight method in general population, health authority and hospital doctors. METHODS: We selected six important elements for each benefit and risk assessment based on previous study. The elements of benefit assessment consisted of disease severity, size of population affected by disease, clinical guidelines recommendation, comparative interventions limitation, improvement of efficacy/effectiveness, improvement of quality of life. The attributes of risk assessment contained overall incidence of adverse events, overall incidence of serious adverse events, discontinuation rate due to adverse events, drug or food interactions, drugs of potential misuse, risk management. 583 subjects constituted 3 groups (general population, health authority and hospital doctors) were selected across the country by quota sampling method and performed survey to evaluate preference of each elements with the swing methods repeatedly. The trained interviewers assisted participant successfully completed survey.   RESULTS: Improvement of efficacy/effectiveness and overall incidence of serious adverse events were revealed as the most important attributes than others for benefit- risk assessment in all three groups. Health authority group outweighed the improvement of efficacy/effectiveness [Mean (±SD): 0.208(±0.04)] and overall incidence of serious adverse events [Mean (±SD): 0.220(±0.05)], while 0.204(±0.03), 0.216(±0.04) in doctor group and 0.197(±0.04), 0.185(±0.04) in general population respectively. In six benefit attributes, the lowest preference score was clinical guidelines recommendation [0.114(±0.04)] in health authority group and [0.144(±0.04)] in general population while size of population affected by disease [0.126(±0.04)] in hospital doctor group. Among six risk elements, the lowest preference was drugs of potential misuse showed in health authority [0.117(±0.04)] and in hospital doctors [0.121(±0.04)] while risk management  [0.121(±0.04)] in general population. CONCLUSIONS: This shows that improvement of efficacy/effectiveness among benefit attributes and overall incidence of serious adverse events among risk attributes are key elements for benefit-risk assessment.