BackgroundThis study developed a short version of a scale measuring attitudes toward people living with dementia, the Four‐Item Attitudes toward People Living with Dementia Scale (APDS4), that could be included in a large population survey.MethodsWe used three datasets from Japan: a web panel survey, a community‐based mail survey, and data from a randomised controlled trial on dementia education. The original scale used was the Attitudes toward People Living with Dementia Scale developed by Kim and Kuroda. Test–retest reliability and item response theory analyses were used to reduce the number of items. The reliability, internal consistency, validity, and responsiveness of the short version were evaluated.ResultsSix items with low test–retest reliability and four items with low discrimination parameters were removed from the 14‐item scale. The APDS4, consisting of four items, had test–retest reliability and internal consistency comparable to those of the original scale. Confirmatory factor analysis indicated that the APDS4 fit a unidimensional model. The validity of the APDS4 was confirmed by significant associations between the APDS4 scores and the original scale scores, knowledge of dementia, helping behaviour intentions toward people living with dementia, helping behaviour experience, attending the Dementia Supporter Training Course, and engagement in healthcare jobs. In a randomised controlled trial dataset, the APDS4 was more responsive to educational interventions than the original scale.ConclusionThe shortened APDS4 was established as a reliable, validated, and responsive scale. This scale can be used efficiently in population surveys to evaluate dementia‐friendly initiatives at the community level.