Objective The IDEAL Schedule was developed for staging “care needs” in patients with dementia. We here aim to validate the Spanish version, further test its psychometric properties and explore a latent construct for “care needs”. Methods A multicenter study was done in 8 dementia care facilities across Spain. Patients referred with a reliable ICD‐10 diagnosis of dementia (n = 151) were assessed with the IDEAL Schedule by pairs of raters. Inter‐rater reliability (intra‐class correlation [ICC] coefficients), internal consistency (Cronbach's alpha), and factor analysis were calculated. Convergent validity for individual items was tested against validated Spanish versions of international instruments. Results Pilot testing with numerical scales supported the feasibility, face, and content validity of the schedule. The psychometric coefficients were good/clinically acceptable: inter‐rater reliability (mean ICC = 0.861; 85% of the ICCs > 0.8), internal consistency (global alpha coefficient = 0.74 in 5 nuclear items), and concurrent validity (global score against the Clinical Dementia Rating schedule, r = 0.63; coefficients for individual items ranging from 0.40 to 0.84, all statistically significant, p < 0.05). Internal consistency was low for the “nonprofessional care” and “social support” dimensions. Factor analysis supported a unidimensional solution, suggesting a latent “care needs” construct. Conclusion The Spanish version of the IDEAL Schedule confirms the main psychometric properties of the original version and documents for the first time the convergent validity of individual items. Factor analysis identified a latent construct consistent with the concept “care needs” although 2 dimensions need further psychometric research.