The aim of this study is to evaluate the validity and reliability of the Mandibular Function Impairment Questionnaire (MFIQ) (Portuguese version). Face and content validity of the Portuguese version were performed. To assess reproducibility of the data gathered with MFIQ, it was applied to 62 individuals who completed the questionnaire on two occasions. Validity and reliability of the data gathered with MFIQ were evaluated in a sample of 249 patients. Construct-related validity was assessed through factorial validity (by means of a confirmatory factor analysis), and convergent and discriminant validities were assessed, respectively, by the average variance extracted (AVE), composite reliability (CC) and bivariate correlations between factors. The internal consistency was estimated by the standardised Cronbach's alpha coefficient (α) and reproducibility by the intra-class correlation coefficient (ICC). All the items of MFIQ showed content validity. Reproducibility was excellent in both the 'functional capacity' dimension (D1) (ICC(D1) =0·895, 95% CI=0·832 to 0·935) and the 'feeding' dimension (D2) (ICC(D2) =0·825, 95% CI= 0·726 to 0·891). Items 1, 2, 6 and 7 of D1 had factor weights below the desired cut-off (0·5), and overall fit of the original bifactorial structure of the MFIQ was poor [(confirmatory fit index) CFI= 0·850, (goodness of fit index) GFI= 0·781, (root mean square error of approximation) RMSEA= 0·118]. Thus, these items were excluded, and the new, reduced version of the MFIQ showed good fit (CFI=0·933, GFI=0·879, RMSEA=0·099). The convergent validity was adequate (AVE≥0·5, CC≥0·7) for both factors. However, their discriminant validity was low (AVE(D1) = 0·51 and AVE(D2) = 0·66 <ρ(2) (D1D2) =0·70). The internal consistency was excellent (α(D1) =0·874; α(D2) =0·918). The Portuguese version of the reduced MFIQ produced data with good validity and reliability.