Background The newly developed Functional Independence and Difficulty Scale is a tool for assessing the performance of basic activities of daily living in terms of both independence and difficulty. The reliability of this new scale has not been assessed. Aims The aim of this study was to examine the relative reliability and absolute reliability of the newly developed scale in community-dwelling frail elderly people in Japan. Methods Participants were 47 community-dwelling elderly subjects (22 for assessing test–retest reliability and 25 for assessing inter-rater reliability). As relative reliability indices, intra-class correlation coefficients were used. From an absolute reliability perspective, we conducted Bland–Altman analysis and calculated the limit of agreement or minimal detectable change to determine the acceptable range of error. Results Intra-class correlation coefficients for test–retest and inter-rater reliability were 0.90 (P < 0.001) and 0.97 (P < 0.001), respectively. The limit of agreement for test–retest reliability was −5.2 to 1.8, representing an increase of over six points for improvement and a decrease of over two points for decline of basic activities of daily living ability. The minimal detectable change for inter-rater reliability was 3.7, indicating that a three-point difference might be existed between difference raters. The results of this study demonstrated that the FIDS appeared to be a reliable instrument for use in Japanese community-dwelling frail elderly people. Conclusions While further research using a large and more diverse sample of participants is needed, our findings support the use of FIDS in clinical practice or clinical research targeting frail elderly Japanese people.