OBJECTIVES:  The aim of our study was to examine the possibilities of the assessment of pain in people with dementia and assess the clinical applicability of these scales in the nursing practice. METHODS: The study was a cross-sectional, descriptive, quantitative analysis. By purposive sampling we enrolled patients with dementia at the age of 60 or older (N=101). Exclusion criteria included not having any other mental or acute somatic disease and not being in terminal condition. The study was carried out at the facility of the Hungarian Baptist Aid in Pécs between 01.12.2015 and 31.01.2016. The analysis was done with Microsoft Office Excel software including descriptive statistics (absolute frequency, relative frequency, mean, standard deviation, confidence interval) and mathematical statistics (Chi-test, T-test ANOVA) at p<0.05. We used five different assessment scales. The NRS (numeric rating scale) is a general one, while the other four scales, PAINAD (Pain Assessment in Advanced Dementia Scale), FLACC (Face, Legs, Activity, Cry, Consolability scale), Doloplus 2 and PACSLAC (Pain Assessment Checklist for Seniors with Limited Ability to Communicate) are special scales made to assess behavior in people with dementia who do not communicate. RESULTS:  We found that the NRS scale indicates excessive sensitivity in people with mild dementia (p=0.03), while in people with severe dementia, the NRS scale did not assess pain so accurately compared to the findings of the other four scales. PAINAD, FLACC, and Doloplus 2 proved to be more reliable. PACSLAC was the most considerable, but its clinical usability seems to be limited due to long and complex evaluation. CONCLUSIONS:  Considering all, we can state that the four dementia-specific scales which rely on patients’ behavior are more reliable in terms of means than the NRS. Using these was more beneficial in our clinical practice.