Abstract This study aimed to assess the reliability and validity of Brazilian-Portuguese versions of the Health Literacy in Dentistry (HeLD) scale in a sample of elderly Brazilian participants. HeLD was initially translated into and cross-culturally adapted to the Brazilian Portuguese language. The reliability and validity of HeLD were then assessed in a sample of 535 non-institutionalized older persons who also completed a questionnaire containing sociodemographic and health information. Data were then randomly separated into two sub-datasets, and Confirmatory Factor Analysis was performed through structural equation modelling, with a maximum likelihood estimate to test the fit of the data to the factor structure of the long-and short-form HeLD (HeLD-29 and HeLD-14) versions of the instrument. The models were compared using the Akaike Information Criterion to assess goodness-of-fit and to determine which models were preferred. Internal consistency of HeLD was evaluated using Cronbach´s coefficient α. Both versions of HeLD were observed to demonstrate high internal reliability (Cronbach´s α ≥ 0.87 for all seven subscales), acceptable convergent (estimates of ≥ 0.50 for AVE and ≥ 0.70 for CR) and discriminant validity. However, the goodness-of-fit of the confirmatory factor analysis models demonstrated satisfactory results only for HeLD-14 subsamples (x2/df = 1.8-2.3; CFI = 0.97-0.98; GFI/NFI = 0.98-0.99; RMSEA = 0.05 and SRMR = 0.03). In conclusion, HeLD-14 was shown to be a reliable and valid instrument to measure oral health literacy in elderly Brazilian participants.