Objectives: To construct a set of quality indicators (QIs) for dementia diagnosis and management in a primary care setting., Design: RAND modified Delphi method, including a postal survey, a stakeholders consensus meeting, a scientific expert consensus meeting, and a demonstration project., Setting: Primary care., Participants: General practitioners (GPs), primary care nurses (PCNs), and informal caregivers (ICs) in postal survey and stakeholders consensus meeting. Eight national dementia experts in scientific consensus meeting. Thirteen GPs in the demonstration project., Measurements: Mean face validity and feasibility scores. Compliance rates using GPs' electronic medical record data., Results: The initial set consisted of 31 QIs. Most indicators showed moderate or good face validity and feasibility scores. Consensus panels reduced the preliminary set used in the demonstration project to 24 QIs. The overall compliance to the QIs was 45.3%. Discriminative validity of the set was good; significant differences in adherence were found between GPs with high and low levels of patients aged 65 and older in their practice, with and without PCNs, and with positive and negative attitudes toward dementia (all P<.05). Based on the demonstration project, one QI was excluded. The final set consisted of 23 QIs; 15 QIs contained innovative quality criteria on collaboration between GPs and PCNs, referral criteria, and assessment of caregivers' needs., Conclusion: This new set of dementia QIs is feasible, reliable, and valid and can be used to improve primary dementia care. Because of the innovative quality criteria, the set is complementary to the existing dementia QIs.