Aim of this study was to perform a comprehensive psychometric analysis of the Italian version of the Oral Health Impact Profile short-form (OHIP-14)--a questionnaire analysing the oral health-related quality of life--using both classical test theory and Rasch analysis. Eighty-five subjects, residents in a long-term nursing home, were analysed using the OHIP-14, Geriatric Oral Health Assessment Instrument (GOHAI) and a global clinical judgement. Two of the five rating categories (1 = almost never; 3 = quite often) did not comply with the Rasch criteria for category functioning. After collapsing rating categories into a three-level rating scale (0 = never; 1 = sometimes; 2 = often), the new model met the set criteria. Item 12 ('difficulty doing usual jobs') was slightly underfitting. At the principal component analysis on the standardized residuals, item 9 ('difficulty to relax') and 11 ('irritable with other people') showed a high correlation (r = 0.66). This indicates that they may not be locally independent. Rasch analysis showed both the substantial unidimensionality of OHIP-14, and the possibility of simplifying the structure of its rating scale. Further studies are needed to confirm the stability of these findings in different testing situations.