BACKGROUND Early detection of dementia may improve patient care and quality of life, yet as many as half of patients are undiagnosed. Electronic health record (EHR) data could potentially be used to help to identify patients at risk of having undiagnosed dementia who could be targeted for outreach and assessment, but acceptability to patients and caregivers is unknown. METHOD We conducted five focus groups at Kaiser Permanente Washington, an integrated healthcare system in Washington State, to explore feelings about timing of dementia diagnosis, use of EHR-33 based tools to predict risk of undiagnosed dementia, and communication about risk. Two team members analyzed transcripts using inductive thematic coding. RESULTS Participant groups included: patients with dementia or mild cognitive impairment, patients with neither diagnosis, and caregivers. People who were non-white or Hispanic were oversampled. Forty patients and caregivers (63% women; 59% non-white/Hispanic) participated in focus 39 groups. Participants supported early diagnosis, describing benefits such as time to adjust to the disease, plan, involve caregivers, and identify resources. They also acknowledged the possible psychosocial toll of receiving the diagnosis. Participants also supported use of an EHR-based tool, but some worried about accuracy and privacy. Participants emphasized that information about dementia risk should be communicated thoughtfully by a trusted provider and should include advice about prognosis, treatment options and other resources. CONCLUSION Overall, patients and caregivers supported using EHR-based tools to help identify patients at risk of having undiagnosed dementia. Such tools must be implemented carefully to address concerns and ensure patients and caregivers are adequately supported.