Background: The Camberwell Assessment of Need for the Elderly (CANE) is widely used for multidimensional evaluation of older people with mental health problems. The aim of this study was to evaluate reliability and validity of a Portuguese version of CANE. Method: A cross-sectional multicenter study was designed using a convenience sample of elderly mental health services' users. CANE was compared with EASYCare, GHQ12, MMSE, Barthel Index and GDS 15 to assess criterion and construct validity. Inter-rater and test-retest reliability were also assessed. Results: 79 patients (76% female), with mean age of 74 ( 6.6) years were included. Most patients lived at home with a family caregiver, generally female. Only 32% had no carer. Dementia was the commonest psychiatric diagnosis (61 %) and somatic comorbidity was very prevalent (85 %). Kappa values (kappa) for interrater item reliability ranged from 0.72 to 1.00 (mean values: 0.96 patient, 0.93 carer, 0.90 staff). The intraclass correlation coefficient (ICC) for total scores ranged from 0.95 to 0.98. For test-retest item reliability, the mean kappa value was: 0.80 patient, 0.77 carer, 0.81 staff. ICC for total scores ranged from 0.82 to 0.92. Given the absence of a gold standard, criterion validity was assessed by comparing CANE with EASYCare (r(s)0.460; p <= 0.01), GDS (r(s)0.615; p <= 0.01), GHQ (r(s)0.581; p <= 0.0 1) and Barthel 4 Index (r(s)-0.435; p <= 0.01). Overall, inter-item and item-total correlations for CANE and item comparison with other measures indicated reasonable construct validity. Conclusion: The psychometric proprieties of CANE seem to be consistently good, in accordance with other studies. Robust results on ecological, face, content, criterion and construct validity, as well as good reliability, were achieved. This version is a promising tool for research and practical use in Portuguese old age settings.